The service works through broker partnership programs. When a client trades under the terms of an eligible partner arrangement, the broker may pay partner compensation based on trading activity that qualifies under that program. ForexLab Rebate then returns a portion of that compensation to the trader. The economic effect is straightforward: part of the effective cost of trading is offset by the rebate.

For algorithmic traders, that can matter more than it first appears. Some automated strategies trade frequently and operate with a relatively small average profit per trade, which makes transaction costs an important part of their overall economics. A rebate does not improve the trading logic, but it can change how much of the strategy’s gross result remains after costs.

ForexLab Rebate is launching as more than a standalone cashback website. It sits alongside EA ForexLab’s independent EA testing, published backtest analysis, Forex EA Database, and a growing set of analytical tools. The rebate business is intended to support the commercial sustainability of that wider research and product ecosystem.

Already trading with a supported broker? Check the available ForexLab Rebate conditions to see whether your existing account can be connected or whether a new account is required.

Commercial disclosure: ForexLab Rebate is a commercial service. EA ForexLab may receive partner compensation from brokers for trading activity generated by users connected under an eligible partnership arrangement. A portion of that compensation may be returned to users as a rebate. Commercial relationships do not determine the conclusions of our independent EA testing. Our testing principles, assumptions and limitations are described in the EA ForexLab Testing Methodology.

What Is a Forex Rebate?

Every trade has a cost. Depending on the broker and account type, that cost may come from the spread, a commission, or a combination of both. On a raw-spread account, a trader may pay a separate commission while receiving a narrower spread. On a standard account, more of the trading cost may be embedded in the spread.

Many brokers also operate partner programs. When a client is connected through an eligible partner arrangement, the broker may share part of the revenue associated with that client’s trading activity, provided the activity qualifies under the program rules. A rebate service returns part of the partner compensation it receives instead of retaining the full amount.

In simplified form:

You trade with a broker → the broker earns revenue from spreads and/or commissions → the broker pays partner compensation on eligible trading activity → the rebate service returns part of that compensation to you.

One technical distinction matters. A forex rebate is not always a literal refund of the exact commission charged on a specific trade. Partner compensation structures vary. Some programs are linked to trading volume, others to spread-derived revenue, and the calculation methodology can differ materially between brokers.

The common economic effect is that the rebate offsets part of the trader’s effective trading costs.

Actual conditions depend on the broker, account type, instrument, trading volume, partner-program rules and, in some cases, jurisdiction. Put simply, not every account type or trade will necessarily qualify. Current conditions should therefore be checked on the relevant broker page.

How Broker Rebate Conditions Can Differ

Forex rebate conditions are not standardized across brokers. Partner compensation can vary by account type, instrument, trading volume and the structure of the individual partner program. That is why ForexLab Rebate will publish the actual client-facing rebate rate, eligibility rules and payout conditions on dedicated broker pages rather than presenting one universal rate.

Important: broker partner programs can change, and some trading activity may be excluded under program, jurisdictional or anti-abuse rules. The current client-facing rebate rate, supported account types, eligibility requirements and payout schedule should always be confirmed on the relevant ForexLab Rebate broker page.

Source note: the examples above describe the general structures found in the broker materials reviewed by EA ForexLab and, where available, publicly stated broker pricing. They do not disclose or guarantee private ForexLab Rebate partner terms. Broker-specific client conditions will be published separately and updated when the underlying arrangements change.

Rebates Improve Trading Economics, Not Strategy Quality

This distinction is important because rebate marketing often blurs it.

A rebate does not create a trading edge and does not repair weak strategy logic. It reduces part of the effective cost of trading. In borderline cases, lower costs can improve the final financial result enough to move a strategy from a small loss to a small profit. That does not mean the strategy itself has become better: its signals, robustness and underlying trading edge remain unchanged.

The real value of a rebate is therefore economic. For strategies that are sensitive to transaction costs, reducing effective expenses can materially improve the net result. Whether the strategy itself has a durable edge remains a separate and more important question.

Why Forex Rebates Can Matter More for Algorithmic Traders

The value of a rebate depends largely on how a specific strategy interacts with trading costs. Relevant factors include trade frequency, average position size, expected profit per trade, instrument, account commission structure, typical spread, execution quality and the terms of the rebate itself.

Some automated strategies sit at the most cost-sensitive end of that spectrum. Short-term and high-turnover EAs may execute many trades while targeting a relatively small average return per position. In such systems, spreads and commissions can consume a meaningful share of the strategy’s gross edge.

Cost sensitivity is therefore one of the factors we examine when testing strategies whose results may be materially affected by spreads, commissions and other trading expenses. It forms part of how EA ForexLab configures and interprets backtests and optimization results.

There is also a practical distinction between trading performance and rebate calculation. In many partner programs, compensation is based on trading activity that qualifies under the program rather than on whether an individual trade was profitable. The exact mechanism varies by broker, but where this structure applies, the rebate functions as a partial offset to trading costs rather than a reward for profitable trading.

What Our Controlled Backtest Simulations Showed

To evaluate the effect of lower effective trading costs in a controlled historical simulation, we ran two comparative backtests in MetaTrader 4. Both tests used historical tick data from Darwinex with Tick Data Suite. Trading-condition parameters were configured using an IC Markets account preset.

In the standard-cost scenario, the commission assumption was set at $7.00 per round-turn lot. In the simulated-rebate scenario, the effective commission was reduced to $5.40 per round-turn lot, representing a modeled rebate effect of $1.60 per lot. The EA logic, settings, market data and trade sequence were otherwise left unchanged.

Important: the $1.60-per-lot figure is a controlled simulation assumption used to isolate the economic effect of lower trading costs. It should not be interpreted as a published or guaranteed ForexLab Rebate rate for IC Markets or any other broker.

The first test used Gold Trend Scalping MT4 5.1 on XAUUSD, M15. Net profit increased from $1,828.47 to $1,867.03, an improvement of $38.56 or 2.11%, across the same 241 completed trades.

The second test used ARS-Trader-EA 2023.12-1 on EURUSD, M5. Net profit increased from $408.75 to $441.07, an improvement of $32.32 or 7.91%, across the same 202 completed trades. A separate analysis of this EA and its test results will be published later.

These are simulations of the economic effect of a lower effective commission assumption. They are not records of historical ForexLab Rebate payouts.

The EAs used in these examples were selected to demonstrate the mechanics of trading-cost reduction. They should not be interpreted as examples of profitable, stable, recommended or otherwise validated trading systems.

The results illustrate the key point: with the trading logic unchanged, lower effective costs can improve the final economic result. The size of that effect depends on trade frequency, position volume, commission structure and the sensitivity of the individual strategy to transaction costs.

For more detail on our testing approach, data assumptions and limitations, see the EA ForexLab Testing Methodology.

Why We Built ForexLab Rebate

EA ForexLab’s core activity is the independent testing and analysis of Expert Advisors. Where appropriate, we use high-quality tick data and Tick Data Suite, model spreads and commissions, account for slippage assumptions, and examine results at the individual-trade level.

The objective is not to create a supposedly “perfect” backtest. It is to make the assumptions more realistic and to understand how sensitive a result is to the conditions under which it was produced. Our broader process is described in the EA ForexLab Testing Methodology.

This work has real costs. Detailed testing requires time, data, software and computing resources. Planned stages such as forward testing on real accounts also require real capital. ForexLab Rebate gives EA ForexLab a commercial revenue stream that is not directly tied to selling individual Expert Advisors, while also helping eligible users offset part of their effective trading costs.

The rebate business should also help us turn years of experience in algorithmic trading research into more useful tools for evaluating trading systems. Our goal is not to declare which strategies “will be profitable.” It is to help traders understand backtest quality, robustness, cost sensitivity and the differences between historical testing and live trading more clearly. Better evidence does not remove uncertainty, but it can make it easier to reject weak systems before real capital is exposed.

The Incentive Problem in a Forex Rebate Model

Every rebate model contains an obvious tension: the service earns money from trading activity, so more volume can mean more revenue. That conflict should be acknowledged rather than hidden.

For EA ForexLab, long-term relationships with users are more valuable than maximizing short-term turnover. A user who takes excessive risk and quickly leaves the market does not create sustainable value for either side. That gives us a rational commercial incentive to build tools and research that help users evaluate trading systems more carefully before committing real capital.

Our interests are not perfectly aligned, and we do not claim that they are. EA ForexLab does not directly profit from a user’s trading profitability, cannot prevent losses, and does not believe that more trading activity is automatically better. Our position is narrower: long-term user value matters more than volume at any cost, and that principle affects the products we choose to build and how we structure the service.

The EA ForexLab Ecosystem for Algorithmic Traders

EA ForexLab is developing an ecosystem around three connected areas:

• Discover & Research — the live Forex EA Database, independent EA testing and published research.
• Analyze & Optimize — Backtest Results Analyzer in development, with ForexLab Portfolio Analyzer and ForexLab Optimization Analyzer planned.
• Validate & Monitor — planned real-account forward-testing infrastructure and public third-party monitoring for selected systems.

ForexLab Rebate sits outside the EA evaluation process as a supporting commercial layer. Its role is to help eligible users offset part of their effective trading costs while contributing to the commercial sustainability of the wider ecosystem.

Why Choose ForexLab Rebate?

We do not claim to offer the highest rebate rates in the market, and we do not claim that ForexLab Rebate is objectively the “best” rebate service. Our proposition is based on advantages that can be demonstrated.

Built Around Algorithmic Trading

EA ForexLab already works with MT4 and MT5 Expert Advisors, backtest analysis, trading metrics and the practical limitations of automated systems. ForexLab Rebate is being developed for the same audience and with the needs of EA users and algorithmic traders in mind.

Connected to an Independent Research Ecosystem

The service exists alongside published EA tests, the Forex EA Database and analytical tools being developed by the same project. This allows trading costs to be viewed as part of the broader evaluation of a strategy rather than as an isolated cashback feature.

A Transparent Commercial Model

We explain where partner compensation comes from, why rebate conditions vary, and how the commercial model supports further development of the wider EA ForexLab ecosystem.

A Broader Analytical Infrastructure

The tools under development build on experience accumulated through years of working with backtests, Expert Advisors and algorithmic trading analysis. The long-term goal is to connect EA discovery, analysis, optimization, forward validation and monitoring in a more coherent research workflow.

What Forex Rebates and EA Testing Cannot Guarantee

Forex rebates cannot guarantee profitability, and trading is never risk-free. A rebate does not repair a fundamentally weak trading strategy; it only improves the economics around that strategy by offsetting part of its trading costs.

A carefully constructed backtest also does not guarantee future performance. The same Expert Advisor can behave very differently in live trading because of execution, spreads, slippage, changing market conditions, broker-specific factors and other variables that historical testing cannot reproduce perfectly. Real-account monitoring shows what has already happened, not what will happen next.

The purpose of better testing is therefore not certainty. Certainty is not available in this field. The goal is to make decisions using more evidence and fewer marketing claims.

How Forex Rebates Work with EA ForexLab

1. Choose a broker from the list of supported partners.
2. Open a new account or connect an existing account where the broker’s partner program allows it.
3. Trade as usual under the broker’s standard trading conditions.
4. Eligible trading activity is calculated according to the rules of the relevant partner program.
5. Receive your rebate according to the payout rules published for that broker and account type.

Broker-specific rebate rates, eligibility rules and payout details will be published and maintained on the relevant ForexLab Rebate broker pages as each partner arrangement is finalized.

The Bigger Vision

EA ForexLab began with a narrow objective: independently test Expert Advisors and publish the results. The wider project grew from a simple principle — traders are better served by evidence, representative testing and transparent methodology than by screenshots and sales pages alone.

Our long-term goal is to build an ecosystem where algorithmic traders can discover systems through the Forex EA Database, examine independent test data and analyze individual backtests more deeply.

Over time, that workflow should also support portfolio-level evaluation, more systematic interpretation of optimization results, comparison of vendor claims with independent and live evidence, clearer risk analysis, and partial compensation of eligible trading costs generated by activity the trader was already undertaking.

Some parts of that ecosystem are already live. Others are in development or remain planned. We will continue to make those distinctions clear as the project evolves.

ForexLab Rebate is one of the mechanisms intended to make that ecosystem commercially sustainable. To explore the current broker options and account requirements, visit ForexLab Rebate.

Risk Disclaimer

Forex trading and the use of automated trading systems involve a high risk of capital loss. Historical results, backtests, forward tests, trading-cost simulations and third-party monitoring do not guarantee future profitability. A rebate may offset part of the effective cost of trading, but it does not remove market risk and should not be treated as a substitute for evaluating the quality and suitability of the underlying strategy.

EA ForexLab content is provided for informational and research purposes only. It does not constitute investment advice, individualized financial advice, or a guarantee of any result. Before trading, you should independently assess the risks involved, the broker’s terms, and whether a particular strategy is appropriate for your circumstances.

Rebate availability, regulatory treatment and any tax consequences may vary by jurisdiction. Users are responsible for checking the rules that apply to them and, where necessary, seeking independent professional advice.

Сообщение Forex Rebates for Algorithmic Traders: Why We Built ForexLab Rebate появились сначала на EA ForexLab.

Key findings first

The tested configuration ran on XAUUSD M15 from January 2024 to early July 2026, on a 100%-real-tick MT5 backtest, $10,000 deposit, fixed 0.01 lot, with EAMode=1. Every figure below is recomputed from the Orders and Deals ledger.

• Net +$90.49 at a profit factor of 1.61 and a 64.64% win rate, reconciled to the cent. On fixed 0.01 lots that is roughly 0.9% over the whole period.
• The strategy works through pending stop orders, and many never fill. It placed 457 stop orders and only 280 became positions, a fill rate near 61%.
• Each filled pending order was submitted with an initial $8.50 SL and $31.64 TP measured from the requested trigger price. All 280 realised exits were sl-labelled; the final stop had moved favourably from its original level in 269 cases, while 11 closed at the original stop. There were no TP-labelled exits.
• The historical expectancy came from winning often, not from winning big. Average win ($1.31) and average loss ($1.29) are near-identical, so the break-even win rate is about 49.6%.
• The trades are extremely short and costs are material: commission and swap took about 32% of the pre-cost result, and a single full stop is worth roughly eight average wins.
• The current product is version 3.20 with two modes, and the public live signals differ from the test in payoff shape, direction and risk. The Mode 1 signal currently reports only 17% Algo trading, which limits how far the complete account history can be read as EA-only evidence.

What I tested

The inputs matter for reading everything that follows. UseFixedLot=true and FixedLot=0.01 match the execution record: all 280 entries were exactly 0.01 lot. The settings also contain MaxConsecLosses=4 and DD_LotMultiplier=0.5; a five-loss streak occurred without any observed change in lot size, but the report does not expose enough internal logic to say why. It also exposes EAMode=1 without an executable version or a definition of the mode, so I treat that as a configuration label whose exact semantics are not independently established. The label is consistent with the vendor’s current Mode 1 naming, but it does not prove configuration equivalence.

The costs reconcile cleanly and the terminology needs care. Summing the Deal Profit field across the ledger gives +$132.37. Take off commission (−$39.54) and swap (−$2.34) and the result lands on the reported +$90.49. That +$132.37 is not the MT5 Gross Profit of $237.71; Gross Profit and Gross Loss are the per-position winning and losing totals, and they net to $90.49 once the platform folds commission back in. Two different statistics, easy to confuse.

How TwisterPro enters the market

The strategy does not send continuous market orders. It places stop-entry orders above or below price and waits for the market to trade through them, and a large share never trigger. Across the test it created 457 pending stop orders, 263 buy stops and 194 sell stops. Of those, 280 filled, 106 expired, and 71 were cancelled, a fill rate of about 61%. Buy stops filled more often than sell stops, 66.9% against 53.6%, which is part of why the executed book leaned long even though the strategy placed orders on both sides.

The cancellations are unusually tidy. All 71 cancelled orders were removed on a Friday at 19:00 server time, with no exceptions, which reads as a weekend housekeeping step that clears resting orders before the close. I am describing what the timestamps show rather than attributing it to a specific named feature in the current build, since the report does not label the mechanism.

How TwisterPro exits, and why the take-profit never mattered

Every filled pending order was submitted with the same geometry relative to its requested trigger price: the initial SL was $8.50 away and the TP $31.64 away, a nominal 3.72:1 ratio. Actual fills did not always occur at the requested trigger, so the distance from the executed entry to the original stop was not fixed: the median was about $8.55, the mean $8.72 and the widest observed distance $15.26. The 3.72:1 figure is therefore order-request geometry, not the realised risk-reward and not even a constant fill-to-stop ratio.

An sl exit is not the same as a losing trade. All 280 realised exits carry an sl label, but the final stop level had moved favourably from the original SL in 269 positions. In the remaining 11, the final stop was unchanged and the position closed at the original stop level. No trade has a TP-labelled exit. The historical result was therefore realised through stop management rather than target hits, but it would be inaccurate to describe every one of the 280 exits as a trailing-stop exit.

Is it really a scalper?

The report puts average holding time at nine minutes, but the distribution is much shorter than that average suggests. The median is about 16.5 seconds. A total of 243 positions closed in under five minutes, and one additional trade closed at exactly 5:00, so 244 of 280 were out in five minutes or less. A small tail of longer positions, with the maximum just under thirteen hours, pulls the arithmetic mean up to nine minutes. On duration, the scalper label fits the observed trade history.

Where the historical expectancy came from

Unlike some breakout systems that win less than half the time but pay off big, TwisterPro is the opposite shape. The average winner and average loser are within two cents of each other, so there is almost no payoff asymmetry to lean on. With a near-symmetric payoff, the break-even win rate sits just under 50%, and the strategy cleared it by roughly fifteen points in this sample. The positive result came from winning more often than the break-even threshold, not from larger winners. That also sets a fragility worth naming: the largest single loss in the record, $10.91, is about eight times the average win, so one full stop that does not trail into profit undoes roughly eight winning trades. A short cluster of full stops would matter more than the smooth curve suggests.

Execution costs and price gaps

Because the edge per trade is small, the cost side is not a footnote. The pre-cost result was $132.37; commission removed $39.54 of it, just under 30%, and swap took a little more. On a per-trade basis the gross edge of about 47 cents dropped to 32 cents net once costs were paid. When the average win is only $1.31, a commission near 14 cents per round turn is a third of the edge.

The tester also lets me measure how far executions landed from the prices the strategy asked for, which is a useful sensitivity check even though it is a simulation rather than a live broker. Comparing each filled order’s requested stop price with its actual fill, the displacement was adverse on 232 of 280 entries, with a median of about five cents, a mean near twenty-two cents, and a worst case of $6.76. Doing the same for the exits, comparing the stop level in each sl comment with the actual close, the median was about four cents, the mean around sixteen cents, and the worst case $5.99. Combined across entry and exit, the median round-trip displacement was about sixteen cents and the mean about thirty-eight cents. These are Strategy Tester requested-versus-executed gaps, not measured live broker slippage, and I keep that label deliberately. What they show is that even in simulation the fills drifted against the strategy, and on an edge this thin that drift is economically relevant. That is consistent with the vendor’s current requirement for a RAW-spread account and its recommendation to run a VPS, but those vendor requirements are a separate matter from the historical cost arithmetic, not something the backtest proves.

Direction and stability

The executed book leaned long, 176 to 104, and both sides won at almost the same rate, so the direction skew did not come with a quality gap between longs and shorts. Splitting the net result by calendar year, every segment was positive: roughly +$16 in 2024, +$40 in 2025, and +$35 in the partial 2026 through early July. I would not read regime information into three annual bars from one backtest, and 2026 is only half a year, but there was no losing year in the sample. Thursday was the only weekday that finished net negative across the whole test, which I note as a descriptive fact about 280 trades rather than a calendar edge worth trading around.

The drawdown behind the smooth curve

The reported 0.28% maximum equity drawdown belongs to this historical simulation with fixed 0.01-lot sizing on a $10,000 deposit. That context matters as much as the percentage itself. Under identical prices and fills, a larger fixed position would mechanically increase dollar gains and losses, while live execution, margin requirements and other nonlinear effects prevent a simple extrapolation of the backtest drawdown. MT5 also reports a Sharpe Ratio of 21.21, but this is the tester’s own metric and should not be read as a directly comparable conventional annualised Sharpe. The curve is smooth; the magnitude of the risk still has to be read together with the sizing.

Historical test versus the current live signals

The product has moved since the tested build. The current MQL5 listing is version 3.20, and the official changelog matters for interpretation: earlier versions reduced the stop-loss distance (v1.8), reworked the mode structure and introduced a shorter-stop Mode 2 (v2.0), added SL recalculation after slipped pending-order fills (v2.1), expanded execution and broker handling (v2.2), redesigned trailing and rejected-exit handling (v2.4), gave Mode 1 selectable Short Stop and Long Stop variants (v2.6), and later reintroduced a prop-firm minimum-hold delay (v3.20). The report exposes no executable version, and a backtest ending in July 2026 could have been produced by a later build, so I cannot map the tested configuration to a specific release. The current product should not be assumed to behave identically to this historical record.

The vendor runs two public live signals, and they are more useful as trading-fingerprint context than as headline-return comparisons. The public Mode 1 signal currently has 69 trades, an 86.95% win rate and a profit factor of 5.08. Its payoff shape differs from the backtest: the average winner is $5.70 while the average loser is $7.49, so the higher hit rate carries more of the result. Average holding time is about one minute, the book is 61% short, and maximal balance drawdown is 5.76%. Those figures come from a different live account, sizing scheme and software history; they are not directly comparable with the fixed-0.01 backtest and do not establish proportional scalability.

There is a further attribution limit on Mode 1. MQL5 currently reports 17% Algo trading on that account. I did not find an authoritative MetaQuotes definition specific to this percentage, so I treat it as a platform-reported attribution signal rather than a precise trade-level classification. Without trade-level attribution, the complete Mode 1 history should not be presented as an EA-only forward test of TwisterPro. The public Mode 2 signal currently reports 100% Algo trading, but it is a different, newer short-stop profile: 51 trades, a 78.43% win rate, PF 2.06, average winner $1.91, average loser $3.37 and average holding time of 28 seconds. It is useful current context, not validation of the tested EAMode=1 configuration. Both signal pages also carry MQL5’s warning about high risk of negative slippage when copying deals.

What the evidence does and does not establish

A few limits are worth stating plainly. The tested version is unknown, so the backtest cannot be mapped onto the current v3.20 or either named mode with certainty. The five-layer validation and internal mode logic are vendor descriptions the order ledger cannot confirm. What the ledger does confirm is one fixed-size position at a time, entered through stop orders, with every realised exit SL-labelled; 269 final stop levels moved favourably from the original SL and 11 did not. The reported 0.28% drawdown is specific to this fixed-0.01-lot simulation on a $10,000 deposit. The profitability also rests on 280 trades and a per-trade expectancy small enough that execution quality remains a material part of the result.

What the test establishes is a coherent, low-exposure scalper with a positive historical expectancy on XAUUSD over this period, driven mainly by a win rate above the break-even threshold in this sample, whose realised risk and return both depend heavily on how it is sized and executed. Whether that is worth $399 and a RAW-spread setup is a decision for the reader; the purpose here is to show what the record actually contains.

If you want to see how we separate a clean backtest from live behaviour, our testing methodology and our note on why forward testing matters set out the approach, and you can compare this record against other XAUUSD systems in the Forex EA database.

A note on costs, since they are central to this strategy: a forex rebate may reduce part of eligible spread or commission across the many small trades a scalper places. It does not reduce slippage, change the entry logic, improve the win rate, or lower market risk, so treat it as an efficiency on transaction costs, not a change to the strategy.

Frequently asked questions

Is TwisterPro Scalper profitable in the EA ForexLab backtest?

Yes, in the tested configuration. The MT5 report shows +$90.49 net over 2024 to 2026 from a $10,000 deposit, at a 1.61 profit factor and a 64.64% win rate, and those figures reconcile against the raw deal ledger. On fixed 0.01 lots that is about 0.9% over the period, so the return should be interpreted together with the tested position size rather than as a general return profile for the EA.

Does TwisterPro use grid or martingale?

Not in this record. Every one of the 280 entries was 0.01 lot, with no lot progression and no averaging into losing positions, and the account never held more than one position at a time. The conclusion applies to the tested configuration and history.

How does TwisterPro enter and exit trades?

It places pending stop orders and only some fill: 457 stop orders produced 280 positions in the test, a fill rate near 61%. Each filled order was submitted with an SL $8.50 and TP $31.64 from the requested trigger price. All 280 realised exits were SL-labelled; the final stop had moved favourably from the original SL in 269 cases, while 11 closed at the original stop. There were no TP-labelled exits.

How long does TwisterPro hold trades?

Very briefly. The median holding time is about 16.5 seconds. 243 of 280 trades closed in under five minutes, and one additional trade closed at exactly 5:00. A small tail of longer positions, up to about thirteen hours, lifts the reported average to nine minutes.

How sensitive is TwisterPro to spread and slippage?

Materially. Commission and swap consumed roughly a third of the pre-cost result, the trades are extremely short, and in the tester the fills moved adversely from the requested prices by a median of about four to five cents on each side. Those are simulated requested-versus-executed gaps rather than measured live broker slippage. They make execution quality economically relevant. Separately, the current vendor requires a RAW-spread account and strongly recommends a VPS.

Does the live TwisterPro signal confirm the backtest?

Not directly. The Mode 1 live signal is still a small sample, shows a higher win rate but a different payoff shape, leans short where the backtest leaned long, and spans a different software and account context. MQL5 also currently reports only 17% Algo trading on that account, so the full history cannot be assumed to be EA-only. It is vendor-associated context, not an independent forward test of the tested configuration.

The archive with the EA files and full test results is available on our Telegram channel EA ForexLAB:

Сообщение TwisterPro Scalper EA MT5 Review: XAUUSD Backtest & Execution Analysis появились сначала на EA ForexLab.

Key findings first

The tested historical configuration ran on XAUUSD H1 from January 2024 to May 2026 on a 100%-real-tick MT5 backtest, $10,000 deposit, fixed 0.01 lot. Every figure below is recalculated from the Orders and Deals ledger.

• Net +$4,574.69, profit factor 1.70, and a win rate of just 46.81%. It made money with more losing trades than winning ones.
• The edge is payoff asymmetry: the average winner (+$12.06) was close to twice the average loser (−$6.10), which puts the break-even win rate near 33.6%.
• Every one of the 1,961 entries was 0.01 lot. There is no lot ladder, no averaging into losers, nothing that behaves like grid or martingale recovery in this record.
• Five labelled breakout modules (A–E) fire the trades, but their realised exposure was often synchronized: 74.6% of entries shared an exact timestamp and price with at least one other entry.
• Grouping exposure from flat to flat gives 669 episodes. The ones that reached four or five simultaneous positions produced about +$4,894, more than the entire account result, while the shallow one and two-position episodes lost money in aggregate.
• The current product is version 3.0 with three distinct modes and live signals that behave very differently from one another, so the tested history is not a clean stand-in for what is on sale now.

What I tested

The costs reconcile cleanly, and the terminology is worth getting right because it is easy to trip over. Summing the Deal Profit field across the ledger gives +$4,962.28. Take off commission (−$268.85) and swap (−$118.74) and you land on the reported +$4,574.69, so costs trimmed about 7.8% off the pre-cost trading result. That +$4,962.28 figure is not the MT5 Gross Profit of $11,073.82; Gross Profit counts only the winning side, while the $4,962.28 is the net of all deals before the separately reported costs. Different statistics, easily confused.

One settings note, since it matters later. The report runs fixed 0.01 lots with Inp_LotMode=0, and several risk inputs read zero (Inp_MaxDrawdownPct=0.0, Inp_DailyDDPct=0.0). I am not going to read a specific meaning into those zeros, because the report does not document what a zero does internally. What I can say is what the ledger shows the EA actually did.

How Gold House actually opens trades

The 46.81% win rate is only part of the picture. I grouped the 1,961 positions into flat-to-flat exposure episodes, tracking the account from the first position of a sequence until exposure returned to zero. That reconstruction shows how the five labelled modules combined at account level rather than treating every position as an independent decision.

Five modules that usually move together

Every entry carries a comment tagging it to one of five modules, from Gold House_PendingA to Gold House_PendingE. On raw counts they look like five separate engines.

The marketing describes five independent breakout strategies, and at the level of code they may well be distinct. Their realised market exposure was not. When I grouped entries by exact timestamp and price, 1,462 of the 1,961 entries, or 74.6%, opened at the same instant and the same price as at least one other entry. The common shapes were four modules firing together, often A, C, D and E, and sometimes all five. Whatever differs inside the five modules, their realised exposure in this historical sample was frequently synchronized rather than diversified: several modules tended to trigger on the same breakout.

Fixed size and maximum observed concurrency

Because several modules can be active together, the natural question is how much exposure the account actually carried. Reconstructing active positions from the deal chronology, the maximum was five open at once, all at 0.01 lot, for a peak aggregate of roughly 0.05 lot. Long and short exposure never overlapped in the record. On this evidence the “no grid, no martingale” description fits the record rather than just being repeated: sizing is flat at 0.01 throughout, there is no progression into losing positions, and the multiple positions come from different breakout modules rather than from averaging down. I found no martingale lot progression or conventional recovery-grid structure in this historical execution record, and I would describe the tested configuration as multi-module breakout stacking rather than recovery trading.

Where the historical profit came from

Grouping the 669 episodes by how many positions they carried at once turns the smooth curve into something more specific, and this is the finding I weigh most heavily.

The one and two-position episodes, which together make up half of all episodes, lost money in aggregate. Episodes that reached four or five simultaneous positions generated about +$2,729 and +$2,165 respectively, roughly +$4,894 combined against a final net of +$4,575. The five-position group closed positive 83.8% of the time. I would not turn that into “more positions cause more profit.” One plausible explanation is that the same price movement both triggered more modules and produced more favourable follow-through, but the number of simultaneous positions does not establish causality. The narrower finding is that historical profitability was concentrated in episodes that reached four or five simultaneous positions.

Fewer winners, larger wins

At the single-trade level, Gold House lost more often than it won, with 918 winners against 1,043 losers. The average winner was roughly twice the average loser, which puts the break-even win rate near 33.6%, below the observed 46.81%. In this historical sample, that payoff asymmetry was sufficient to offset a win rate below 50%. The result therefore depended on larger winners compensating for more frequent losses, which is consistent with periods of relatively flat equity between the stronger winning sequences.

How Gold House exits trades

Exit labels are where a breakout system’s real behaviour shows, and they need reading carefully. Of the 1,961 exits, 1,886 carry an sl comment, 18 carry tp, and 57 have no SL or TP label. An sl exit is not automatically a loss: many of these are stops that were trailed into profit and then hit, which is consistent with the trailing-stop behaviour the vendor describes, though the ledger shows the effect rather than proving the mechanism. On raw Deal Profit, the sl-labelled exits actually sum to about +$2,867, the 18 tp exits to about +$1,437, and the rest to about +$658.

A small take-profit tail does a lot of work

Those 18 take-profit exits are worth isolating. They are only 0.9% of all positions, every one of them is profitable, and they average around $80 each, which is many times the $12 average winner. A handful of trend-extension trades that ran all the way to a take-profit target contributed a disproportionate slice of the gross winnings. That fits the payoff-asymmetry picture: the system tolerates a lot of small stop-outs to stay in the game for the rare position that travels a long way.

Profit against maximum favourable excursion (top) and maximum adverse excursion (bottom). Realised profit tracks favourable excursion far more tightly than adverse excursion.

The report’s excursion data points the same way. Realised profit correlates strongly with each trade’s maximum favourable excursion (about 0.91) and much more weakly with maximum adverse excursion (about 0.42). In plain terms, realised profit was strongly associated with how far a trade eventually ran in its favour, and only weakly with how far it first went offside. That association is consistent with a “let winners run, cap losers” design, but a correlation is not proof of the underlying rule, so I read it as a historical association rather than a mechanism.

The remaining 57 unlabelled exits turned out to be a rare case where an input name and the ledger line up: 54 of them occurred on a Friday at 23:00, which matches the configured weekend-close behaviour (Inp_WeekendClose=true, with a Friday close hour set in the inputs). That is the kind of empirical confirmation I trust, because it connects a setting to an observed execution rather than assuming what the setting does.

Holding time and trading frequency

The report puts average holding time at about two and a half hours, with the shortest position lasting seven seconds and the longest just under three days. The distribution is front-loaded in the minutes-to-hours range with a thin multi-hour-to-multi-day tail. I would not call this a scalper despite the many short trades, nor a pure swing system despite the tail; the vendor’s “swing breakout” label is a reasonable fit for a system that takes lots of quick breakout attempts and occasionally holds one for a real move.

The drawdown behind the smooth curve

The 2.25% maximum relative equity drawdown is the number that makes this backtest look remarkable, and it deserves a careful caveat rather than applause. It is the largest equity dip in this particular historical simulation, on fixed 0.01 lots, over this specific 2024–2026 gold market. It is not a structural guarantee. Two things about the architecture argue for caution: the profit depended on episodes stacking four and five positions at once, and the strategy leans on rare large winners, both of which can behave differently in a market that trends less cleanly or whipsaws through breakout levels. The live signals, which I turn to next, show materially larger drawdowns than 2.25%, which is the clearest sign that the backtest figure is specific to its configuration and sample.

Historical test versus the current Gold House

Here the evidence has to be handled with care, because “Gold House” today is not one behaviour. The product is on version 3.0, and the vendor offers it in three modes, so a single historical backtest cannot validate all of it.

Version 3.0 changed the product

The tested report does not expose a version number, and the data ending in May 2026 does not tell me which build produced it, since a backtest over older data can be run on newer software. The input set looks compatible with the pre-v3 classic controls, but I treat that as a hypothesis, not a fact. What is documented is that v3.0, released in mid-2026, added an Adaptive Mode and reworked the exit and management logic while preserving compatibility with the earlier behaviour. The current product is offered in three modes, so it spans several behaviour profiles, and the backtest speaks only to the fixed-lot, multi-module configuration it actually ran.

Three live signals, three profiles

The vendor runs three public MQL5 signals, and the Profit Priority account is the closest match to the tested payoff fingerprint. At the 13 August 2026 snapshot, 43.56% of its 528 trades were profitable, while the average winner was +$14.79 and the average loser −$7.82. Its best and worst trades, +$126.89 and −$27.57, were also close to the backtest’s +$126.51 and −$27.73. That similarity is a useful cross-check, but it is not exact forward validation. The live account’s average holding time is much shorter, its maximal balance drawdown is 10.48% rather than the backtest’s 2.25% equity drawdown, and MQL5 reports that 80% of its growth was achieved within seven days.

The other two signals have materially different payoff profiles. BE Priority currently wins about 73% of trades, with an average winner slightly smaller than the average loser and a profit factor of 2.35. The Adaptive Mode account is explicitly labelled by the vendor as a high-risk configuration reference, not a recommended setting; at the 13 August 2026 snapshot it was down 40.32% with a maximal balance drawdown of 79.33%. That account should not be generalised to Gold House as a whole. The three signals are better treated as separate operating profiles, with Profit Priority the closest match to the tested payoff shape.

What the evidence does not establish

A few limits worth stating plainly. The tested version is unknown, so I cannot map the backtest onto v2.1, v3.0 or the current Adaptive Mode. The five module labels are exactly that, labels; the ledger shows synchronized exposure, not five proven-independent strategies, and it cannot confirm what each module’s internal logic is. The strong profit-versus-favourable-excursion correlation describes the winners without proving the trailing rule that produced them. And the live signals are vendor-controlled monitoring that spans multiple product versions, so they are context rather than an independent forward test. These are limits on what one backtest and a set of public signals can show, not judgements about the EA.

Conclusion

Gold House MT5, in the configuration I tested, is a fixed-lot XAUUSD breakout system that stacks up to five simultaneous 0.01-lot positions from five labelled modules and makes its money from payoff asymmetry rather than a high win rate. It lost on most individual trades and still returned +$4,574.69 at a 1.70 profit factor, because its winners were about twice its losers and because a small number of trend-extension trades ran a long way. The reconstruction shows that most historical profit came from episodes where four or five modules were exposed on the same breakout; the shallow episodes were slightly negative in aggregate. The reported 2.25% drawdown is specific to this fixed-lot backtest and sample, while the public live signals currently show materially larger drawdowns.

Each tested position used only 0.01 lot, but account-level exposure could rise to five simultaneous positions when several modules aligned, so the question worth asking is how that stacked exposure behaves, and how a strategy that leans on rare large winners performs across market conditions other than the 2024–2026 sample. Anyone evaluating the current product should also keep the version gap in view: v3.0 offers three modes that behave very differently, and only one live signal resembles the tested payoff shape. Test the specific mode and settings you intend to run, on demo, before committing. It is worth comparing the structure against other XAUUSD systems in our Forex EA database, and reading our testing methodology and note on why forward testing matters for how we separate a clean backtest from live behaviour.

A note on costs, since this multi-module breakout system trades often: a forex rebate may reduce part of eligible spread or commission across repeated trades. It does not reduce drawdown, swap, or the simultaneous exposure described above, so treat it as an efficiency on costs, not a change to the strategy’s risk.

The archive with the EA files and full test results is available on our Telegram channel:

Frequently asked questions

Is Gold House MT5 profitable in the EA ForexLab backtest?

Yes, in the tested historical configuration. The MT5 report shows +$4,574.69 net over 2024–2026 from a $10,000 deposit, a 1.70 profit factor and a 46.81% win rate, and those figures reconcile against the raw deal ledger. It is a simulation, not a live result.

Does Gold House use martingale or grid trading?

No martingale lot progression or conventional recovery-grid structure appears in this historical execution record. Every one of the 1,961 entries was 0.01 lot, with no lot progression and no averaging into losing positions. The multiple simultaneous positions come from five separate breakout modules, and maximum concurrency was five at 0.05 lot aggregate. The conclusion applies to the tested configuration and history.

Why is Gold House profitable with a win rate below 50%?

Payoff asymmetry. The average winner (+$12.06) was close to twice the average loser (−$6.10), which puts the break-even win rate near 33.6%. Winning 46.81% with that payoff ratio was sufficient for positive historical expectancy, with a small tail of larger winners contributing materially to the result.

Does the Gold House live signal match the backtest?

Partly. The Profit Priority signal preserves the tested payoff fingerprint, a sub-50% win rate with winners roughly twice the losers and near-identical best and worst trades. But its holding time is shorter and its drawdown larger than the backtest, and the other two signals (BE Priority, Adaptive High-Risk) behave very differently, so it is context rather than exact forward validation.

Which Gold House version was tested?

Unknown. The report does not expose a version field, and a backtest over data ending May 2026 could have been run on later software. The current product is version 3.0 with three modes, so the tested fixed-lot configuration should not be assumed identical to what is on sale now.

What is the maximum drawdown in the historical test?

2.25% maximum relative equity drawdown, and 1.95% by balance. That is specific to this fixed-lot backtest and sample; the vendor’s live signals show materially larger drawdowns, including one high-risk reference account near 79%.

Сообщение Gold House MT5 Review: XAUUSD Backtest & Live Signal Analysis появились сначала на EA ForexLab.

Key findings first

The test I examined is the FXStabilizer_EUR 1.2 configuration on EURUSD H1, run from January 2021 to February 2026 at 99.90% modelling quality from a $1,000 deposit. Every figure below reconciles against the raw order table.

• Net profit +$1,678.33, profit factor 1.36, position win rate 59.92%, relative drawdown 54.03%.
• The 756 executed positions reconstruct into 366 flat-to-flat cycles, and 362 of them closed positive. That 98.9% is the share of reconstructed flat-to-flat cycles that ended with positive displayed P/L, not a trade win rate, and the two answer different questions.
• The system recovers a losing position by adding same-direction trades. The first recovery entry stays at 0.01 lot; from the step after that the size doubles, up to 0.64, and cumulative open volume reaches about 1.28 lots at maximum depth.
• Only two cycles ever reached the maximum depth of eight positions. Both were SELL sequences, both were closed by a common stop, and each realised about −$1,013.
• Those two events account for roughly 99.6% of all negative cycle-level P/L, and they more than explain the only two negative years in the sample.
• The vendor’s live Myfxbook accounts share several aggregate traits with the tested run, but they do not establish that the same recovery logic was used, so I treat them as context rather than validation.

What I tested

Before trusting any of the headline figures, I reconciled them against the order rows. Summing the profit on every closed trade gives $1,678.11, which is $0.22 short of the reported $1,678.33. The gap splits between gross profit (off by $0.13) and gross loss (off by $0.09), while the running balance column ends at exactly $2,678.33. The $0.22 difference is consistent with cumulative display rounding across the two-decimal trade rows, so I treat the MT4 headline and final balance as authoritative. Everything else ties out: profit factor 1.36, expected payoff $2.22, the 381 long and 375 short positions, the largest winner and loser.

Two things about the test itself shape how I read it. It is a simulation. The report uses Every Tick modelling, shows 99.90% modelling quality, zero mismatched-chart errors and variable spread, but it still cannot reproduce every feature of live execution. The input reading Slippage=20 is an EA parameter, not evidence that twenty points of realistic slippage were modelled. Anyone weighing a live deployment should keep our backtest testing methodology in mind here, because the gap between a clean tester run and real execution is exactly where recovery systems get tested. The expected payoff of $2.22 is already an after-cost number. The question worth asking is not whether costs erased it, but how the structure turned a fairly ordinary win rate into a positive expectancy at all.

What 756 trades actually represent

By direction the trade list looks unremarkable: 381 long, 375 short, winning 59.92% of the time, which reads like an even two-sided strategy. The headline win rate was not the part that interested me. Once I grouped the 756 positions into flat-to-flat exposure cycles, tracking account exposure from the first position of a sequence until every position in it closed and exposure returned to zero, the structure changed considerably. The 756 trades collapse into 366 cycles. Most ended at the first position. The deeper cycles explain most of the risk.

How a losing position becomes a recovery ladder

The order record lays out the mechanism. A cycle opens with a 0.01-lot market order and places a same-direction pending limit one step away, also at 0.01. If the first trade reaches its target, the pending is deleted and the cycle ends clean. If price runs against the position, that first recovery fills at 0.01, a new same-direction limit is placed one step deeper, and the take-profits of the open positions are pulled together toward a common basket exit. From that next step onward the added size doubles: 0.02, 0.04, 0.08, and so on. The sequence keeps adding until the basket closes in aggregate profit, which most of the time it does.

The size progression is where the exposure becomes obvious. The observed ladder runs 0.01, 0.01, 0.02, 0.04, 0.08, 0.16, 0.32, 0.64, and the size distribution across the whole test confirms it: 547 positions at 0.01 down to just two at 0.64. A cycle that reaches the eighth rung is holding about 1.28 lots in aggregate, roughly 128 times the initial entry. That figure is cumulative open volume rather than a risk multiplier, since the positions sit at different prices and the stops move, but it is the honest way to describe how far the exposure can stretch when a sequence runs to its limit.

Why 98.9% of cycles closing positive is not a 98.9% win rate

This is the distinction that reframes the test. At the position level, 40% of trades lose. At the cycle level, only four of 366 sequences finished negative, so 98.9% closed positive. Both numbers are correct, and they describe different things. The recovery structure folds individual losing legs into baskets that, once the converged take-profit triggers, close green overall. Calling 98.9% a win rate would blur two different units of analysis. It simply means that 362 of 366 reconstructed flat-to-flat cycles ended with positive displayed P/L. Of the four negative cycles, one small depth-three loss of $4.96 was created by the tester’s end-of-sample forced close rather than a normal strategy exit. The more important question is what happened in the genuinely failed recovery sequences.

Where the historical profit came from

Grouping the cycles by how deep they ran shows where the money was made and lost.

More than half of the cycles ended at the first position, and progressively fewer reached the deeper recovery levels. The P/L distribution across depth is less intuitive. In this sample, cycles ending at depths one through seven were positive in aggregate, while the two cycles that reached depth eight lost about $2,024 combined. Almost all negative cycle-level P/L is concentrated in that final depth.

Long, short, and where the shortfall really sits

By direction the split is instructive. The 181 buy cycles netted about +$1,832, while the 185 sell cycles netted about −$154, which makes the short side look like the weak leg. Remove the two maximum-depth SELL failures, though, and the remaining 183 short cycles made roughly +$1,870. Two sequences were large enough to turn the whole short-side result from strongly positive to slightly negative. This is a concentration observation, not a claim about how the strategy would perform without its worst trades. You cannot keep the good cycles and delete the terminal ones, because the same logic produces both. The point is narrower: the aggregate short-side figure is dominated by two events rather than by broad weakness.

The two maximum-depth failures

Two sequences changed how I read this test, one in March 2025 and one in January 2026. Both reached the maximum observed depth of eight positions. Both were shorts. Both ended the same way, and that shared terminal pattern is why they matter more than any headline ratio.

In March 2025 the system began shorting EURUSD at 1.04503 and kept adding as the pair climbed, position by position, up through 1.08096. This is the price path that pushed the tested recovery sequence to its maximum observed depth: a sustained one-way move against the basket, with no pullback deep enough to let the converged take-profit trigger. By the eighth position the account held about 1.28 lots short into a rising market. January 2026 repeats the pattern almost beat for beat, shorting from 1.16622 up to 1.20197 on the same ladder and the same depth.

What the order log shows at the eighth position

The modifications just before the end show how the terminal risk control behaved in this tested build. Through the first seven positions, each older trade carries a distant stop that is pushed further away as the basket deepens, so the system is holding on and waiting for the reversal. Then the eighth position, the 0.64-lot order, opens with a tight stop-loss placed just beyond its own entry. In the same instant, the order log modifies the stop-loss of all seven earlier positions to that identical level. A few hours later price touches it, and all eight positions close together at one price.

I will describe that carefully, because the wording matters. In the tested v1.2 history, reaching the eighth position was followed by a common broker-side stop being applied across the entire open sequence, after which the whole basket closed at once. That is what the execution record shows. Whether it corresponds to a named “hard drawdown control” feature in the current product is a separate question I return to below. In March 2025 the common stop sat at 1.08395 and realised −$1,013.02; in January 2026 it sat at 1.20490 and realised −$1,011.29. In balance terms, each of those single events took roughly a quarter of the account, about 25% in 2025 and 27.5% in 2026.

One more distinction is worth keeping straight. Those two cycles are about 99.6% of the negative cycle-level P/L, which is not the same as the MT4 Gross Loss of −$4,724.13. Gross Loss adds up every losing individual trade, including losing legs inside cycles that eventually finished positive. Cycle-level negative P/L counts only sequences whose total result was negative. The first sums losses on individual positions, including losing legs inside cycles that later finished positive; the second counts only cycles whose total result was negative.

The drawdown behind the smooth curve

The report gives three drawdown numbers, and they are not interchangeable. Absolute drawdown ($465.86) measures the dip below the starting deposit. Maximal drawdown ($1,386.43, 34.27%) is the largest money fall from a peak. Relative drawdown (54.03%, $627.73) is the largest percentage fall, measured at a different and smaller-balance moment, which is why the dollar figure attached to the 54% is smaller than the one attached to the 34%. Neither should be quietly replaced by a gentler realised number reconstructed from closed balances, because a closed-order reconstruction cannot see intra-basket floating equity, and the report’s own equity-based figures stand.

What the reconstruction does explain is where the visible cliffs come from. The two near-vertical drops late in the curve, one in early 2025 and one in early 2026, coincide with the two depth-eight stop-outs, and the volume sub-window shows its largest spikes underneath them. For four years the curve is a quiet diagonal because the recovery mechanism kept resolving. It stops being quiet in precisely the two places the mechanism ran to its limit.

What changed in 2025 and 2026

Grouped by the year a position was entered, the sample runs positive every year from 2021 to 2024, then turns negative in 2025 and the partial 2026. It is tempting to read that as the strategy breaking down, but the reconstruction points somewhere more specific. Take the single March 2025 event out of that year and the rest of 2025 contributed roughly +$749; the maximum-depth loss more than explains the negative annual figure on its own. The partial 2026 sample tells a milder version of the same story: outside the January event, the year was close to flat, near +$25. These are loss-concentration diagnostics, not an alternative strategy result, and two observations are only two observations. What they show is that the negative years were not caused by broad deterioration across ordinary trades. Each was defined by one terminal recovery failure of about a thousand dollars.

How long FXStabilizer holds its trades

This is not a fast system, whatever the smooth curve suggests. Rebuilt from actual entry-to-close times, the median position was held about 48 hours, the mean nearly four days, and the longest single position stayed open more than 40 days. Deeper cycles hold longer, which is what you would expect from a mechanism that waits for a converged basket target. That carries a cost the MT4 summary table does not itemise. The historical report exposes no separate swap column, so I will not invent one, but positions open for days or weeks accrue overnight financing, and any real-account version of this result is sensitive to it. The vendor’s own live accounts make the point, since both carry sizeable negative interest, which I come to next.

Vendor backtests and the recent test answer different questions

FXStabilizer’s official site publishes its own EURUSD backtests, and it is worth being clear about why they and this test do not contradict each other. The official Durable run covers 2009–2016; the Turbo run covers 2013–2016. Both use AutoRisk=true with different risk limits, a $10,000 deposit, fixed spread and 90% modelling quality on a different demo server. The configuration I tested uses Mode=0, AutoRisk=false, RiskLimit=50, a $1,000 deposit and variable spread at 99.90% quality over 2021–2026. The vendor backtests show how earlier risk-scaled configurations behaved on quote history ending in 2016; this test shows how a fixed-lot configuration behaved over the last five years, a period the official pages do not cover. The useful difference is the period: the EA ForexLab test covers 2021–2026, which the official historical pages do not. I would not call the older backtests invalid for being old, and I would not read the newer one as a verdict on the vendor’s published numbers, because they do not overlap.

What the live Myfxbook accounts do and don’t show

The vendor points to two long-running Myfxbook accounts for EURUSD. Both are useful as long as they are read for what they are, which is vendor-associated monitoring rather than independent tests.

Turbo EURUSD, live, and a familiar profile

The Turbo EURUSD account is a real Markets4you cent account on MT4, still updating when I checked it, showing a large multi-year gain against a 13.26% drawdown. Its aggregate profile looks a lot like the tested configuration. Long and short win rates sit in the low-to-mid 60s, the average winner is near $14.87 and the average loser near $15.29 against $14.13 and $15.59 in the backtest, and the reported average trade length is about one day. It also carries about −$1,420 of accumulated interest, which is the financing cost I flagged earlier showing up on a real account.

Durable EURUSD, and reading the zero balance correctly

The Durable EURUSD account needs care. Its balance reads $0.00, which looks alarming until you check the ledger: deposits of $10,000, withdrawals of $117,701.18, and a peak balance of exactly $117,701.18. The account did not blow up. The entire balance was withdrawn, which matches the vendor’s note that FXOpen discontinued its STP account type and funds were moved out rather than lost. It last updated in January 2025. As a live record it shows the same family traits: EURUSD on MT4, directional win rates in the high-50s to low-60s, four-day average holds, and about −$10,500 of accumulated interest over its life.

Does the live trading match the tested configuration?

Only in part, and only in the parts I can actually measure. The public monitoring shares several aggregate characteristics with the tested configuration, including EURUSD trading on MT4, broadly similar directional win rates, an asymmetric average winner and loser profile, and holding periods measured in about one to several days. Those similarities are useful context, but they stop at aggregate statistics. What the public statistics do not establish is that the same recovery ladder, the same lot progression, or the same maximum-depth exit logic was used, because Myfxbook shows aggregate results, not the order-by-order structure I reconstructed from the tester report. The versions, modes, brokers, leverage and deposits also differ from the tested v1.2_fix, and neither account discloses its settings. The accounts therefore provide limited live context. They do not confirm that this specific configuration, recovery ladder or maximum-depth logic produced a particular live result. That is why I treat forward testing as a separate evidence layer rather than an extension of the backtest.

What this test does not establish

A few things I deliberately will not claim, because the evidence does not support them. The tested inputs read Mode=0, and the current product sells Durable and Turbo modes, but the official backtests carry no Mode parameter and use AutoRisk=true, so I cannot map Mode=0 onto Durable or Turbo without documentation I do not have. The current Ultimate version advertises “hard drawdown control,” and the tested history clearly applies a common hard stop at maximum depth, but I am describing an execution pattern in one build, not confirming it is that named feature. The vendor’s line that its security mechanism has not triggered “since 1997” refers to a different pair and configuration than the one here; in this EURUSD test the terminal stop plainly did trigger, twice. Version lineage across builds, and the mapping of inputs to marketed modes, remain open questions. For a second example of how recovery exposure can change the interpretation of a smooth backtest, see our Opal EA recovery-grid analysis.

Conclusion

Stripped to its structure, FXStabilizer_EUR v1.2 is a two-sided EURUSD recovery system that often closes the first position profitably and, when that entry moves against it, adds same-direction positions on a progressive lot ladder in an attempt to bring the sequence back to a shared exit. Over 2021–2026 that produced a smooth-looking +$1,678.33 at a 1.36 profit factor, with 362 of 366 cycles closing positive. The reconstruction is what gives that curve its texture. Cycles ending at depths one through seven were positive in aggregate, while risk was concentrated almost entirely in the two cycles that reached depth eight. Each of those events reduced realised balance by roughly a quarter when a sustained EURUSD move pushed the ladder to its terminal stop.

The risk characteristics visible in the evidence are straightforward to name: progressive multi-position exposure that can reach roughly 1.28 lots on a small account; financing costs on multi-day holds that the summary does not itemise; and a terminal outcome that, in this sample, arrived twice and defined both losing years. The trade-off is specific and measurable: a high share of positive reconstructed cycles alongside rare, sharp terminal losses at maximum observed depth. Anyone weighing it should look past the diagonal equity line, decide how they feel about that trade-off, and test their intended mode and broker on demo with financing and maximum-depth exposure in full view. Our Forex EA database is a reasonable place to compare it against alternatives before committing real money.

On costs specifically: a forex rebate may reduce part of eligible spread or commission across repeated multi-position recovery sequences. It does not reduce swap, it does not reduce drawdown, and it does nothing to change the maximum-depth exposure described above, so treat it as a small efficiency, not a safety margin.

Frequently asked questions

Is FXStabilizer a grid or martingale system?

The tested execution record shows a grid-like recovery structure with progressive position sizing. The first recovery entry stays at 0.01 lot, followed by an observed doubling sequence up to 0.64 lot at maximum depth, with positions closed as a basket. What the record establishes is that behaviour; it does not reveal the internal design intent behind it, so I would describe it by what it does rather than force it into a single label.

Did the backtest make money?

Yes. The MT4 report shows +$1,678.33 net over 2021–2026 from a $1,000 deposit, a 1.36 profit factor and a 59.92% position win rate. The sum of the displayed trade rows is $0.22 lower, a difference consistent with cumulative display rounding. It is a simulation, not a live result.

How can 98.9% of cycles be positive if 40% of trades lose?

They measure different things. Individual trades lose 40% of the time, but the recovery structure folds losing legs into baskets that usually close positive overall, which happened in 362 of 366 reconstructed cycles. The 98.9% simply means that 362 of 366 reconstructed flat-to-flat cycles ended with positive displayed P/L; it is not the odds on a single trade.

What actually caused the losses?

Two sequences, in March 2025 and January 2026. Both shorted EURUSD into a sustained rise, added positions to the eighth rung of the ladder, and were closed by a common stop for about −$1,013 each. Together they are roughly 99.6% of all negative cycle-level P/L, and each more than explains its year’s negative result.

Is the vendor’s Myfxbook proof it works live?

It is supporting context, not proof of this configuration. The live accounts are vendor-associated, run different modes, brokers and deposits than the tested v1.2_fix, and one of them (Durable) stopped updating in January 2025 after its balance was fully withdrawn. They share some aggregate characteristics with the backtest, but Myfxbook does not expose the order-level structure, so it cannot confirm the same recovery ladder was used.

Did the Durable account blow up?

No. Its balance shows $0.00 because the full balance was withdrawn, with withdrawals of $117,701.18 equal to its peak, consistent with the broker closing its STP account type. That is an account closure and cash-out, not a trading loss.

The archive with the EA files and full test results is available on our Telegram channel:

Сообщение FXStabilizer EA Review: EURUSD Backtest, Recovery Risk & Drawdown появились сначала на EA ForexLab.

Key findings first

The tested version is ORION GOLD Scalping 2.0, run on XAUUSD H1 from 2024.01.01 to 2026.05.25 in MetaTrader 5 with 100% real-tick modelling and a $10,000 starting deposit. The headline P/L and the structural figures below reconcile against the raw deal history.

• Net +$7,368.81, profit factor 1.43, win rate 65.64%, maximum relative equity drawdown 20.94%.
• All 946 reported positions were long. There was not a single short entry in the record.
• Those 946 positions reconstruct into just 157 flat-to-flat cycles, and none of them finished after a single entry. Every cycle grew to between 2 and 9 positions.
• Of the 789 entries added after each cycle’s first trade, 642 were opened higher than the previous entry and only 147 lower.
• Commission and swap together reduced pre-cost profit by 24.7%; swap accounted for most of that cost.
• The test ended below its March 2026 peak. The final balance sat $3,284.06 under the high, exactly matching the reported maximum balance drawdown.
• The Myfxbook account often cited alongside this EA is a v4.0 MetaTrader 4 demo, not the v2.0 I tested, and it’s a good lesson in why a “100% closed winners” line tells you less than it seems.

What I tested

Here’s what I was working with:

• Expert: ORION GOLD Scalping 2.0 (MagicNumber 123123).
• Platform / server: MetaTrader 5, RannForex-Server, Build 5833.
• Symbol / timeframe: XAUUSD, H1.
• Period: 2024.01.01 to 2026.05.25 (the last trade closed 2026.05.20).
• Modelling: every tick based on real ticks, quality 100% (432,230,066 ticks, 14,514 bars).
• Deposit / leverage: $10,000, 1:500.

A backtest is still a simulation. A 100% real-tick model improves the historical reconstruction, but it is not evidence of live execution and I do not read it as a forward result. That distinction is part of the EA ForexLab testing methodology. The useful part here is not the profit number by itself; it is the complete deal history, because that lets us inspect how the EA traded.

Two configured inputs come back later. The settings carry a DailyLossLimit of 492.0, plus two very permissive caps: MaxGridOrders of 70 and MaxOpenLotSize of 3.0. As it turned out, the EA never came close to either cap, so those numbers describe what it was allowed to do, not what it did.

What ORION GOLD actually did

The supplied MT5 results block and balance curve: +$7,368.81 net, PF 1.43, 946 trades.

Every position was long

When I split the deals by direction, the picture was clean: 946 buy entries, 946 sell exits and zero sell entries. In MetaTrader terms, each “sell out” is the closing side of a long, not a short. The tested execution record is therefore 100% long. In this run the EA opened long positions in gold only; the sell deals are exits.

946 trades are really 157 cycles

MT5 reports 946 trades, all of them long. Treating those 946 positions as 946 independent decisions is misleading. Once I reconstructed the account from flat exposure back to flat exposure, they collapsed into only 157 cycles. Not one cycle finished after a single entry. Every cycle expanded to at least two positions, the median depth was seven, and in 147 of 157 cycles all positions closed at the same timestamp, consistent with synchronized multi-position exits rather than independent trades.

That changes how I read the headline statistic. MT5’s 65.64% win rate is a position-level number, while the economically useful unit in this reconstruction is the full exposure cycle. The two are not interchangeable.

The lot ladder

Inside a cycle the size steps up on a repeating ladder: 0.01, 0.01, 0.01, 0.02, 0.03, 0.05, 0.07, 0.10, 0.14 lots. The biggest single position in the run was 0.14 lot, while the largest simultaneous exposure was 0.44 lot across nine positions. That is 44 times the initial 0.01-lot clip in cumulative volume. Even so, the observed run stayed far below the configured 3.0-lot and 70-order caps, so those caps were never tested by this sample.

Recovery grid or pyramiding?

Multi-entry gold EAs are often discussed as either averaging-down grids or pyramiding systems, but ORION does not fit neatly into either label. I went in expecting a fairly ordinary recovery grid. The entry sequence did not support that simple reading. This is the same distinction that matters when comparing grid and martingale EAs: the label matters less than the actual exposure path.

Most additions were higher, not lower

Across the 157 cycles there were 789 entries added after the first. When I separated the ones opened above the previous entry from the ones opened below, the split was lopsided: 642 higher, 147 lower, none equal. A system that mostly adds at higher prices isn’t a classic averaging-down grid. To be precise about what that is and isn’t: I’m reading the price at which each entry filled, not the intent behind it. The report tells me where the EA added, not why its code chose to.

Entry-path classification

Sorting each cycle by its price path is where the result got interesting.

The asymmetry is hard to miss. The all-higher group, where every addition was opened above the previous entry, produced more than the account’s final net profit. The other two path groups were negative over the sample. Put plainly, historical profitability was concentrated in cycles whose successive entries were made at progressively higher XAUUSD prices. That is an association in this dataset, not proof of a particular internal module.

The depth-7 versus depth-8 gap

Group the cycles by depth and an odd gap appears: all fifteen depth-7 cycles were negative, while the sixty-three depth-8 cycles were strongly positive. It would be easy to read that as evidence that deeper cycles worked better, but the composition changes at the same time. The depth-8 group is almost entirely all-higher; the losing depth-7 cluster is dominated by mixed and lower paths. In this sample, entry-path composition tracks the depth-7/depth-8 difference more closely than depth alone. Depth by itself is therefore a poor summary of what separated those groups.

Where the historical result came from

MT5 reported an expected payoff of $7.79 per trade after costs. Commission and swap are already inside that figure. To see what they changed, I reconciled them separately: together they reduced the sum of the Deal Profit field before those costs by 24.7%. That is large enough to matter when interpreting the historical result.

Before those two line items, the per-deal profit field summed to +$9,789.96. Commission took −$396.77 and swap took −$2,024.38, leaving the reported +$7,368.81. In this test, swap was roughly five times the commission. Nine cycles were positive on the Deal Profit field but negative after costs, and in every one of those cases swap was the component that changed the sign.

Order holding duration and weekly trading results ORION GOLD Scalping

I would not carry that number unchanged to another broker. In this RannForex environment, prolonged long exposure produced a substantial negative swap cost, but XAUUSD financing varies by broker, entity and period. The defensible conclusion is narrower: financing was material to this historical run and should be included when judging any multi-day implementation of the strategy.

The 2026 slide

Split by the year each cycle started, the account made money in 2024 and 2025 and lost a little across the partial 2026 sample. What’s worth noticing is which behaviour slipped. The all-higher group stayed positive in every year, 2026 included. Through 2024 and 2025, its gains comfortably outweighed the losses from the other two path types; in the 2026 stub that balance tipped the other way, as the lower-price and mixed cycles lost more than the all-higher cycles brought in.

Drawdown, and where the test stopped

Two drawdown figures get conflated all the time, so it’s worth keeping them apart. Maximum balance drawdown ($3,284.06, 15.90%) is the largest fall on closed balance. Maximum equity drawdown ($3,381.03, 16.33%) and relative equity drawdown (20.94%, $2,392.30) include open, floating positions and are measured at different peak-to-trough events. They’re not in conflict; they answer different questions.

One detail from the reconstruction stuck with me. Closed balance peaked at $20,652.87 on 16 March 2026 and finished the sample at $17,368.81, exactly $3,284.06 lower. That is the reported maximum balance drawdown. The test therefore ended without recovering that realised-balance decline. This says something about where the sample stopped, not what would have happened next, but it belongs beside the headline profit when judging the result.

Is it really a scalper?

The name suggests fast in-and-out trades. The report shows a different exposure profile: average holding time was 19 hours 11 minutes, and the longest position stayed open for 235 hours 29 minutes, nearly ten days. There are short-duration trades in the sample, but plenty of exposure lasted for hours or days. That matters because those longer holds are also where financing costs become relevant.

What Myfxbook does and doesn’t prove

A Myfxbook account is often circulated with this EA: the public ForexEALab page titled ORION GOLD SCALPER V4.0_fix. It is useful context, but it is not forward validation of the v2.0 I tested. The account name resembles the reviewing site’s brand; I am not making an ownership claim from that name and treat the page purely as a public demo record. This is also why forward testing only validates the configuration that is actually being run.

What the page actually shows is a Demo account on MetaTrader 4 at JustMarkets, 1:500, joined 25 February 2026, magic 783261. That is a different platform, broker and magic number from the tested MT5 v2.0. The page was still last updated on 9 July 2026 when I checked it again on 12 August. At that recorded update it showed +576.92% gain, 84.67% drawdown and a $6,807.87 balance from a $1,000 deposit.

Why “100% closed winners” is the wrong comfort

Myfxbook’s Advanced Statistics panel displays 1,696 trades and marks 113 of 113 longs and 1,583 of 1,583 shorts as winners. The same panel notes that its statistics reference the analysed history and the last 200 transactions, so I treat this as Myfxbook’s displayed summary rather than a reconstructed ledger. Read alone, the winner counts look exceptional. The rest of the page changes the picture. Drawdown reached 84.67%, and the last recorded snapshot showed ten open positions: four buys from 17 June and six sells from 30 June, with a combined floating loss of −$2,060.11, about −30%. No stop-loss was visible in that snapshot.

This is the part of the Myfxbook page I would pay most attention to. Closed win rate describes realised trades; equity and drawdown also reflect unresolved exposure. A strategy can therefore show a perfect closed record while losing positions remain open. That is exactly why 100% closed winners and 84.67% drawdown are not contradictory. None of this makes the monitoring fake, and the floating losses do not have to become realised losses. It simply means closed win rate is an incomplete risk measure.

One more difference matters: this demo trades both directions and is mostly short, whereas the tested v2.0 record is long-only. That alone is enough to stop treating the two records as equivalent.

What changes with the current V5?

The product currently presented on the official Orion site is ORION GOLD SCALPER V5.0 for MetaTrader 5 and XAUUSD. The vendor describes it as rule-based automation, not AI, with configurable risk profiles, session filters, directional modes (Auto, Buy-only, Sell-only), Recovery Control and other protection settings.

So there are three generations in play, and they don’t collapse into each other:

The code lineage between the tested v2.0 executable, the later v4.0 demo monitoring and the current V5 product could not be independently established. They differ in version, platform, direction capability and monitoring status, while reseller pages use several different version numbers. What I tested is the v2.0 executable. I am not extending its evidence to the other builds.

What V5 claims, and why this test doesn’t validate it

Keeping current claims separate from the v2.0 evidence, V5 is presented as rule-based MT5 automation with Recovery Control, calculated SL/TP and configurable protection. The official site says recovery actions can increase exposure, and that a configured position, daily-loss or account-loss limit may pause new trades or request an exit depending on the selected rule. Those are claims about V5. My v2.0 backtest neither confirms nor contradicts them, because it is a different build and the code lineage is unproven. The official site publishes no account-level performance figures for V5, so there is no current vendor performance record to reconcile here.

A note on the daily loss limit

The v2.0 report configured a DailyLossLimit of 492.0, and a cluster of losing cycles closed within a few dollars of roughly −$500. That clustering is consistent with the control affecting some exits, but the supplied report does not establish the internal meaning of DailyLossAction. I therefore would not call it a “$492 stop-loss” or claim that it definitely closed those cycles. The number is suggestive, not proof of a mechanism.

Conclusion

Stripped of the marketing, the ORION GOLD Scalping 2.0 I tested is a long-only, multi-entry gold system that reconstructed into 157 cycles of 2–9 positions on a stepped lot ladder. Historical profit was concentrated in cycles that added successive positions at progressively higher prices. Over 2024–2026 the backtest netted +$7,368.81 at a 1.43 profit factor. It was a profitable historical result, but the per-trade payoff, swap cost and unrecovered March 2026 balance drawdown materially change how I read it.

The reconstruction is solid, but the limitations matter just as much. This is one backtest, one build and one broker’s data, with idealised fills. The Myfxbook demo is a different version on a different platform; its perfect closed record sits beside an 84.67% drawdown and roughly 30% floating loss in the last recorded snapshot. The current product is another version again, with no published account performance and no established lineage back to the tested executable. None of those risks is visible in the headline profit or win-rate figures. If I were evaluating ORION today, I would test the current build separately on demo and watch financing costs, cumulative exposure and open equity rather than the closed win rate alone.

One practical footnote on costs: a forex rebate may offset part of eligible spread or commission. It does not reduce swap, drawdown or the strategy’s exposure. In this test swap was the larger explicit cost, so a rebate would only address a small part of the overall cost picture.

For context against other independently tested systems, see the Forex EA Database.

Frequently asked questions

Is ORION GOLD Scalper a grid or a martingale?

On the tested v2.0 record it builds multi-entry cycles of 2–9 positions on a stepped lot ladder, which is grid-like in structure. But most additions were made at higher prices, not lower, so it doesn’t behave like a classic averaging-down martingale. The historical profit was concentrated in cycles whose entries stepped progressively higher.

Did the backtest actually make money?

Yes. The backtest produced +$7,368.81 net over 2024–2026 from a $10,000 deposit, at a 1.43 profit factor and a 65.64% win rate. Those figures reconcile against the raw deal history. It remains a simulation on one broker’s data, not a live result.

Is the Myfxbook page proof it works live?

No. That account is a MetaTrader 4 demo of a different version (v4.0_fix) on a different broker, and it was still last updated on 9 July 2026 when I rechecked it on 12 August. It also showed a large drawdown and roughly 30% floating loss on open positions. That is useful context, not live validation of the tested v2.0.

How can a “100% win rate” sit next to an 84% drawdown?

Win rate counts only closed trades. Open losing positions are absent from that count but still affect equity and drawdown. So a 100% closed win rate can coexist with a severe equity drawdown; the two metrics are measuring different parts of the account.

Is the current version the same as what was tested?

The current product is ORION GOLD SCALPER V5.0 (MT5). The tested build was v2.0; the demo was v4.0 (MT4). The lineage between the three couldn’t be independently established, so I wouldn’t assume results from one carry to another.

What risk is missing from the headline numbers?

Two things stand out: progressive multi-position exposure and the cost of holding long gold positions for days. On the tested run, swap alone absorbed about a fifth of the pre-cost Deal Profit field.

You can download the archive containing the tests and Expert Advisor files from our Telegram channel:

Сообщение ORION GOLD Scalper Review: MT5 Backtest, Trade Structure & Myfxbook Results появились сначала на EA ForexLab.

The order history explains the structure. Most cycles opened one position, targeted roughly a pip and a half, and closed within minutes. When an entry moved the wrong way, Opal added same-direction positions on a widening grid, scaled the volume, and moved the group toward one shared exit. I reconstructed 6,009 of those cycles from the raw order rows; 88% never needed a second position.

Two results from that reconstruction are worth stating up front. The reported 21.78% drawdown is not a closed-balance figure — the closed-trade balance never fell more than 6.10% — and the evidence points to a single eight-position basket held over the 2023/24 New Year as its main source. Separately, the vendor’s round-number claim is testable against the trade record, and initial entries show strong avoidance around whole-figure levels.

Key findings

• Profitable, with a thin margin. Net $574.95, Profit Factor 1.34, expected payoff $0.08 per position. Break-even win rate implied by the average payoff is 81.76%; observed was 85.74%.
• Around $548.75 of non-price deductions are inferred from the difference between executed-price P/L ($1,123.46) and booked P/L. About $517.58 reconciles to commission under the tested account schedule.
• 81.4% of net profit came from cycles that never needed a second position. Cycles that did invoke recovery remained profitable in aggregate (+$106.83), but the deepest groups became far less productive.
• No broker-side Stop Loss appears anywhere in the record — zero non-zero S/L across all 16,584 entry, modification and exit rows — and every exit is a take-profit, including legs that closed at a loss inside a basket.
• The observed volume progression matches 0.01 × 1.27ⁿ rounded to the broker step, giving a 0.01 / 0.01 / 0.02 / 0.02 / 0.03 / 0.03 / 0.04 / 0.05 ladder and 0.21 lots at maximum basket exposure.
• Accessible live statistics are broadly consistent with the tested profile on win rate, holding time and payoff shape, but only ten rows from the 1,972-item live history were retrievable, so no basket-by-basket comparison was possible.

What was tested

Opal is an MT4 Expert Advisor sold on the MQL5 Market by Oeyvind Borgsoe, who is also listed as CEO of Relevant Trading. The product page was published on 11 November 2023 and shows version 1.0, priced at $580. The stated recommendations are EURUSD, H1, a $100 minimum deposit, and an ECN or RAW account on a VPS, with the default configuration described by the vendor as low-mid risk. The description lists a Money Management module, round level price identification, a time filter, a news protection filter and a recovery system.

The question this test was built around is not whether the historical curve rises. It is what the system does when the first position is wrong, since that is where a recovery grid concentrates its risk.

Test setup and headline results

Test setup

The test ran on an IC Markets demo server, MT4 build 1470, every-tick model at 99.90% modelling quality with zero mismatched chart errors, across 37,623 hourly bars and 201 million modelled ticks. Historical tick data and variable spread came from Tick Data Suite with Dukascopy data, following the EA ForexLab testing methodology.

The report prints no commission line, but a non-price deduction is recoverable from the results. A 0.01-lot position closing 1.5 pips in profit corresponds to $0.15 of price movement; the modal booked win is $0.08. Measuring the difference between executed-price P/L and booked P/L across intraday positions gives a median of $0.07 per 0.01 lot, scaling linearly to $0.14 at 0.02 lots and $0.21 at 0.03 lots — $7.00 per round-turn standard lot, matching the IC Markets Raw Spread commission schedule. Slippage was not separately simulated. MaxSpread=1.5 is an EA filter input and is not a measurement of the historical spread.

MT4 backtest results

Every figure below was checked against the report and recomputed from the 16,584 order rows underneath it.

Summing the displayed profit column gives $574.71 against the report’s $574.95 summary — a difference of $0.24 across 7,033 rows, consistent with displayed-row rounding or hidden terminal precision. The MT4 summary is used for headline figures; reconstructed rows are used for everything derived.

Why the 85.7% win rate needs context

With an average winner of $0.3752 and an average loser of $1.6821, the break-even win rate is 1.6821 ÷ (1.6821 + 0.3752) = 81.76%. Opal delivered 85.74%, a margin of 3.98 percentage points.

That framing is arithmetically correct but uses the wrong unit for this system. Many of the 7,033 positions are not independent outcomes — they are legs of the same trade decision, opened minutes or days apart and closed at the same instant. Grouping them changes what the statistics describe.

How Opal’s recovery system works

How the recovery system is structured

Positions were grouped into cycles by direction and shared exit timestamp, and the grouping was then validated rather than assumed. All 711 multi-position groups close at a single common timestamp, and all 711 carry a single common modified take-profit across every member — a property independent of the grouping rule itself. The 2,518 order-modification rows are the mechanism: recovery positions open with no take-profit or with one subsequently rewritten, and the group’s take-profit levels are aligned to one shared price.

Most observed additions cluster tightly around the configured Step1=15, Step2=20 and Step=40 thresholds, while a small number of large spacing outliers require separate interpretation. Median leg-2 spacing was 15.1 pips, median leg-3 spacing 20.1 pips, and median leg-4-onward spacing 40.0–40.1 pips, with 98.5%, 94.6% and 95–100% of observations within one pip of threshold. Only six of the 1,024 recovery additions were spaced more than five pips beyond their threshold; each was audited individually. Two were leg-2 additions on a single fast hour (25.6 and 28.3 pips). Two crossed multi-day adverse moves before the next fill (135.3 and 136.8 pips, one over a weekend), which are legitimate large overshoots rather than threshold breaches. The remaining two belong to one basket whose three entries share a single timestamp, where the apparent 35-pip spacing is an artefact of ordering three simultaneous fills — measured between price-adjacent levels, that basket’s spacing is exactly 15 and 20 pips. No outlier indicated a different recovery behaviour.

The observed volume progression is consistent with a geometric rule. Scale=1.27 applied compounding and rounded to the 0.01 broker step gives 0.01, 0.01, 0.02, 0.02, 0.03, 0.03, 0.04, 0.05, which matches the realised sequence in all three depth-8 baskets and the modal lot at every leg index — though the trade history shows the progression rather than exposing the internal rule. Across 7,033 positions, 6,720 were 0.01 lots, 276 were 0.02, 29 were 0.03, five were 0.04 and three were 0.05.

The label matters here. A martingale doubling from 0.01 would reach 1.28 lots at the eighth step and 2.55 lots cumulatively. Opal reaches 0.05 and 0.21 — a 21-fold increase in exposure rather than 255-fold. Progressive recovery grid describes the geometry more accurately than martingale, and the operative risk is cumulative exposure held open during an adverse move rather than the label itself.

How often recovery was invoked

Of 6,009 reconstructed cycles, 5,298 closed on the first position — 88.17%. The remaining 11.83% expanded.

Maximum same-direction depth was eight, matching MaxOrders=8. Total account concurrency is a separate figure: buy and sell sequences carry separate magic numbers (BMagic=1, SMagic=2), both RepeatBuy and RepeatSell were enabled, and the two can coexist. The observed peak was ten simultaneous positions on 15 March 2023 at 20:05, with an eight-position buy basket alongside two shorts. Both directions were open simultaneously for 0.71% of the test.

The observation that neither direction exceeded eight while the account reached ten is consistent with MaxOrders constraining a sequence rather than the account. That reading is an interpretation of observed behaviour and is not confirmed by product documentation. Separately, a cap on further additions is not itself a loss cap.

Where the profit was concentrated

Aggregating final basket P/L by depth:

• Single-position cycles: 5,298 cycles, +$467.88 (81.4% of reconstructed net)
• Two to four positions: 691 cycles, +$116.93
• Five to eight positions: 20 cycles, −$10.10

A second decomposition by leg index within the multi-position cycles cuts against the obvious reading of that table. Inside those 711 baskets, the initial positions realised −$1,018.61 and the added recovery legs realised +$1,125.44, netting +$106.83. Only one of the 711 initial legs closed positive; 72.07% of the 1,024 recovery legs did.

Those leg-level figures do not isolate a counterfactual. The realised exit price of every leg in a basket — including the first — is set by the basket-level take-profit that the recovery process itself produced. What the initial position would have returned had no further position been opened cannot be answered from the order history, because the exit that actually occurred is a product of the recovery mechanism. Constructing a no-recovery scenario from hindsight prices would not be evidence.

What the distribution supports is narrower: about 81% of reconstructed net profit came from cycles that never needed a second position; cycles that invoked recovery remained profitable in aggregate; and the deepest groups became much less productive. That shows where the historical profit was concentrated. It does not isolate the marginal contribution of the recovery mechanism itself.

The deepest recovery basket

One sell basket opened on 11 December 2023 and closed on 2 January 2024 — 521 hours, 21.7 days, eight positions.

This basket contains both the largest losing position in the test (−$24.12) and the largest winning position (+$52.56). The first two entries lost $46.73 between them; the last two returned $78.61. The group closed +$6.66.

All eight rows are labelled t/p, including the four that lost money. The take-profit closing the group is a basket-level price, and positions on the wrong side of it are closed at a loss by the same order. “Everything closes at take-profit” describes the order type, not the outcome.

The February 2022 outlier

The most profitable cycle in six years was a four-position sell basket opened on Friday 25 February 2022 between 14:00 and 18:16, at 1.11850, 1.12004, 1.12204 and 1.12605. It closed at 1.11398 for +$50.58 — 8.80% of net profit from one basket.

The four sell legs above are the entire basket. The basket closed at 00:05 server time on Monday, immediately after the weekend interval in the tested history, at 1.11398 — roughly 100 pips below the highest entry. Whether the closing move was a genuine weekend re-opening gap is not established from the trade record alone, but the group resolved across the weekend break rather than through continuous intraday movement.

Excluding this basket, the test returns $524.13. That is a concentration measure, not an alternative backtest — removing the best outcome from a sample is not a fair test of anything. The relevant point is that a single basket held at a loss contributed 8.8% of six years of net profit, resolving across a discontinuity the strategy cannot schedule.

The worst basket, and the largest losing trade

The largest single losing position, −$24.12, belonged to a basket that finished in profit. The worst basket was a different event: a seven-position buy sequence opened on 27 July 2023 and closed eight days later on 4 August for −$9.14, with legs ranging from −$17.29 to +$31.39.

58 of 6,009 cycles finished negative, averaging −$0.71. At cycle level the Profit Factor is 15.05 and 98.98% of cycles are positive. That figure is an architectural property of basket systems rather than a performance claim: a structure that resolves almost everything positively concentrates its risk in the rare cases where it does not.

What changed the outcome of the deepest baskets

Among the 20 baskets that reached five positions or more, all eleven sell baskets closed positive and all nine buy baskets closed negative. Decomposing each basket into executed-price P/L, inferred commission and a residual non-commission adjustment shows where the split came from.

The adjustment is defined so that a negative value is a deduction from booked P/L and a positive value is a credit: booked P/L minus (executed-price P/L minus inferred commission). The confirmed derived result is arithmetic. For all nine deep buy baskets, executed-price P/L minus inferred commission remained positive while the booked result was negative, so a residual non-commission adjustment changed the sign in nine of nine cases. Across the nine that adjustment totalled −$31.95. On the sell side it ran the other way, a net +$13.00 credit across eleven baskets.

The interpretation, which the report does not let me confirm: the size of that adjustment ($1.11 to $9.57 per basket, far larger than rounding), its dependence on multi-day holding, and its long-deduction / short-credit asymmetry are a pattern consistent with asymmetric swap or carry. The MT4 report prints no separate swap column, and the swap schedule applied by the tester is not documented in the test record, so I describe this as a residual non-commission adjustment rather than a reconstructed swap figure. The arithmetic is what stands: a residual non-commission adjustment changed the sign of every deep buy basket in this sample, each of which had positive price movement.

Stop loss and loss limits

All 7,033 orders carry S/L = 0.00000, and all 7,033 closed at take-profit. There were no stop-outs and no equity-triggered exits in the test.

No standard broker-side Stop Loss was recorded — not on any of the 7,033 entries, and not on any of the 2,518 later order-modification records either, so no stop was added or moved into place after entry. The stronger finding holds: non-zero broker-side S/L observations across every entry and modify record number zero. The configured equity-protection switch was disabled in this test — EquitySave=false, so EquityRisk=70 was not active as an enabled protection setting. The order history does not establish what additional internal exit logic, if any, could apply after a basket reaches the observed maximum depth. On the three occasions a basket reached eight positions, it resolved at the shared take-profit.

Filters and entry logic

Time filter

All 6,009 new cycles opened in hours 03 through 20 server time, consistent with StartHour=3 and StopHour=21, with none outside. All 6,009 also opened at minute 00, consistent with a decision taken on the H1 bar open.

Recovery additions behave differently: 58 of 1,024 fell outside the window, spanning 21:00 through 02:00, and only 4% landed on the hour. The pattern is consistent with the time filter restricting new cycle initiation while leaving an existing basket under management. Server time is used throughout; no timezone conversion is asserted.

News and year-end protection

News filter. UseNewsFilter was true, with low, middle, high and NFP pause windows configured, and the chart panel displays “News: No Effective News!”. The historical test does not establish whether a usable historical news dataset was available to the EA during the run. The backtest therefore does not validate the effectiveness of the news-protection feature. I did not build an economic-calendar event-matching test — that would require a documented calendar source, event impact classification, a broker-server-to-UTC mapping and defined matching windows — so I make no claim here about trading during specific events, in either direction.

Year-end protection. The record here splits in two. Across every year of the test, no new cycle opened between 1 and 15 January. The first new cycle of each year fell on 16 January 2020, 19 January 2021, 18 January 2022, 16 January 2023, 16 January 2024, 20 January 2025 and 19 January 2026 — a repeatable suppression matching the span implied by XDAfterEY=15.

The other half is not visible. Ten new cycles opened on 31 December across the test, and December trading continued at normal daily rates through the configured pre-year-end window. Two baskets carried across the year boundary, including the eight-position December 2023 basket, which added recovery positions on 18, 19, 22 and 27 December and closed on 2 January. Two further recovery additions were made on 3 January 2022 to a basket opened on 31 December 2021.

The order history does not demonstrate the pre-year-end restriction or the close-out behaviour implied by CloseEndYear operating as the parameter names suggest, while the post-year-end suppression is present in every year. Parameter semantics may differ from the names, and a tester environment may not exercise every code path.

Round-number levels

The vendor states that Opal uses psychological round-number levels and publishes a separate Round Level Indicator described as drawing psychologically significant levels. The public descriptions do not specify which level family the EA uses, so four families were tested against entry prices: whole figures at 100-pip spacing, 50-pip levels, 20-pip levels and 10-pip levels.

Avoidance appears only around the 100-pip whole-figure family (1.0800, 1.0900 and so on), and only for initial cycle entries. Recovery additions serve as an internal control, since they are generated by a different part of the process.

Among the 6,009 reconstructed initial entries, none occurred within 10 pips below a whole figure, and only ten (0.17%) occurred within 10 pips above one — and those ten all sit at 9.3 to 9.9 pips, at the very edge of the zone. Across the rest of the 100-pip band the entries distribute fairly evenly, between roughly 11.4% and 13.9% per 10-pip slice. The 1,024 recovery additions, generated by a different part of the process, show no such gap: they fall within 10 pips of a whole figure at 10.5% above and 9.7% below, indistinguishable from the rest of the band.

This pattern is strongly consistent with whole-figure avoidance for initial entries. I have stopped short of attaching a significance test to it, because the correct null model is not a uniform distribution: EURUSD did not spend equal time at every price location during the eligible trading hours, and hourly observations are serially dependent, so a naïve test against a flat baseline would overstate the result. The order history contains the entries that occurred but not the full set of eligible bars where no entry occurred, so the market-exposure denominator needed to estimate how much of the gap is attributable to the entry rule cannot be reconstructed from it. The descriptive gap is clear; its exact magnitude relative to market exposure is not quantified here.

The trade history alone does not establish the internal rule, and an input named Rndsize=10 is not proof of its implementation. The effect is also specific to the whole-figure family: at 50-pip and 20-pip spacing no comparable gap appears, so the finding should not be generalised to every “psychological level” the vendor’s wider material references.

Robustness and risk structure

Trade duration

The median position lasted six minutes; the mean lasted three hours and 25 minutes. 47.3% of positions closed within five minutes and 78.2% within an hour, with a tail extending to 21.7 days.

Single-position cycles have a median of two minutes. Positions belonging to a multi-position basket have a median of 124 minutes and a mean of nearly 13 hours. Median duration by depth rises monotonically: two minutes at depth one, 1.7 hours at depth two, 4.7 hours at depth three, 26 hours at depth four, and two to 6.5 days beyond.

Duration does not cause the loss. Baskets remain open longer because price has moved against the position and the shared exit is further away, so the holding time reflects the adverse move rather than producing it.

Performance by year

Every calendar-year segment in the tested sample was positive. The complete years 2021–2025 were all positive; 2020 and 2026 are partial periods, since the test starts on 13 January 2020 and ends on 20 February 2026. The spread across the complete years runs from $53.58 to $178.34, and 2022 contributed 31% of the total, with a single February basket accounting for 28% of that year. Positive annual bars establish consistency of sign rather than consistency of return, and are not on their own evidence of regime robustness.

The chart above groups position results by trade open year, matching the supplied QuantAnalyzer visual; the research table groups complete reconstructed baskets by the year their cycle started. For baskets that open in December and close in January the two conventions assign the result to different years, so the figures differ slightly — 2021 and 2022, for instance, read $55.92 and $176.00 at cycle-start level against $53.58 and $178.34 at position-open level. They answer slightly different questions and are labelled accordingly.

Historical data after the publication date

The MQL5 product page was published on 11 November 2023 and shows version 1.0. Splitting the test at that exact date, with each cycle assigned by its initial entry date, gives the following. No cycle straddles the cutoff, so the split is unambiguous.

• Before 11 November 2023: 4,125 cycles, 4,838 positions, +$401.39, Profit Factor 1.335, win rate 85.66%, 105.7 positions per month
• From 11 November 2023: 1,884 cycles, 2,195 positions, +$173.32, Profit Factor 1.355, win rate 85.92%, 80.7 positions per month

The historical segment after the 11 November 2023 publication date remained profitable, with a similar position-level Profit Factor and win rate, although trade frequency was lower. This is not out-of-sample evidence: it cannot be established that the algorithm was frozen, that later data were never used in optimisation, or that the compiled executable is unchanged. The version number is suggestive but is not proof.

Cycle-level Monte Carlo

A Monte Carlo run at individual-trade level is structurally unsuitable for this system. Positions inside one basket are opened conditionally on the previous ones and closed together, so randomly reordering 7,033 individual trades breaks that dependency and generates sequences the EA could not produce. The QuantAnalyzer Monte Carlo supplied with the test set was not used for any published figure, since its simulation count, resampling method and confidence configuration are not established.

The simulation below treats each of the 6,009 reconstructed baskets as one indivisible unit, so positions inside a basket are never separated. 10,000 permutations and 10,000 bootstrap resamples were run.

One point has to come before the numbers, because the numbers are small and could be read the wrong way. This Monte Carlo is not an estimate of maximum account equity drawdown. It tests only the ordering and mix of realised basket outcomes. The drawdown figures below are closed-basket sequence drawdowns — how far a running total of already-realised basket results dips — and they deliberately exclude the floating exposure inside an open basket, which is the actual risk of this strategy.

• Permutation (order varies, outcome set fixed): final result invariant at $574.71; maximum closed-basket sequence drawdown median $9.14, 95th percentile $10.18, 99th percentile $13.26, worst of 10,000 runs $16.66.
• Bootstrap (order and mix vary): final result median $570.39, 5th percentile $507.57, 1st percentile $491.89, worst run $462.01; maximum closed-basket sequence drawdown 95th percentile $10.55, worst $24.17.

These figures describe sequence risk on a closed-basket series and nothing more. The simulation does not model floating intra-basket equity, new market price paths, spread or slippage changes, regime changes, dependency between cycles, overlapping opposite-direction exposure, margin and stop-out constraints, or any departure from independent and identically distributed conditions. It says nothing about the exposure open on the account in December 2023, or about what a basket would do at a depth the sample never reached.

What the reported drawdown measures

MT4’s Maximal and Relative Drawdown are equity-based metrics that include floating positions, not simply closed-balance drawdown. Reconstructing the closed-trade balance curve across all 7,033 rows gives a deepest balance drawdown of $86.35, or 6.10% — a separately derived metric, not a restatement of the report’s $306.74.

Even that $86.35 reflects booking order: it occurs entirely within the simultaneous close of the December 2023 basket on 2 January 2024, where the profitable legs are booked before the losing ones. Measured basket by basket on realised results, the deepest drawdown across six years is $9.14.

On attribution, the reported $306.74 is consistent with the December 2023 eight-position sell basket. In late December 2023 the only positions open were that basket’s eight legs, totalling 0.21 lots at a volume-weighted entry of 1.09915. The adverse price required to produce a $306.74 floating loss on that exposure is approximately 1.11376, which lies inside the late-December 2023 EURUSD range, and the closed balance immediately before 28 December was $1,409.99 against an equity peak of $1,408.36 implied by the report’s own 21.78% ratio. The timing, basket exposure and required adverse price movement are consistent with that basket being the main source of the reported equity drawdown. This is not an exact reconciliation: it does not incorporate the trough timestamp, the bid/ask side applicable to each position, accumulated swap at that moment or commission treatment, and tick-level equity reconstruction was not performed.

Neither reconstructed figure replaces the report’s 21.78% equity drawdown, and the smaller numbers do not mean the system is safer after aggregation. The basket-series drawdown collapses each recovery basket into one realised observation precisely by excluding the floating exposure that is the strategy’s main risk, and the cycle-level Profit Factor of 15.05 and 98.98% cycle win rate are structural properties of a basket system rather than metrics comparable with the position-level figures. If anything, the distance between the realised numbers and the equity drawdown is the point: the risk lives in the gap.

Live results vs backtest

Myfxbook / SignalStart comparison

Opal has a vendor-controlled live MT4 account at Vantage Markets, running since November 2023 and exposed through both Myfxbook and SignalStart. The two services identify the same underlying Vantage Markets MT4 account and report closely aligned statistics: SignalStart is powered by Myfxbook, the Myfxbook page links directly to the SignalStart listing, both report Real USD accounts at Vantage Markets on MetaTrader 4 at 1:500, and both show Profit Factor 1.82 and average trade length 3h 2m. The small differences between them — 1,966 against 1,970 trades, 2,463.4 against 2,468.3 pips — are consistent with update or synchronisation timing rather than two different accounts. They are one live record visible in two places, not two independent forward tests.

The aggregate live statistics are broadly consistent with the tested configuration in win rate (86% against 85.92% in the post-publication segment), average winner (2.99 pips against 3.01) and average holding time (3h 2m against 3h 58m). Those three come from the monitoring page’s account-wide figures.

The remaining observations come only from the ten accessible history rows and cannot be extended to the full account. In those ten rows, all positions were 0.01 lots, no broker-side Stop Loss was shown, and entries fell just after the hour (17:00:01, 18:00:02, 19:00:01 and similar). Three separate quantities need to be kept apart. Displayed take-profit: seven of the ten rows carry an explicit TP, at roughly 0.5 to 1.6 pips from the open; the other three show no TP value. Realised price movement: all ten rows are positive, ranging from +0.5 to +2.0 pips. Booked P/L: from −$0.01 to +$0.14 — the one trade that gained only +0.5 pip still finished at −$0.01 after the account’s non-price effects, which is the same tight-margin arithmetic the backtest shows on its smallest wins. The displayed live TP distances are broadly compatible with the tested 1.5-pip target, but the sample is small and TP visibility incomplete, and none of the three quantities should be read as a proxy for the others.

They are not sufficient on their own to establish a close basket-by-basket mechanical match. The live trade table is paginated through a scripted interface, and only the ten most recent rows from the 1,972-item history view were retrievable — SignalStart separately reports 1,966 trades, the six-record difference being unidentified account events I have not attributed. The Myfxbook hourly and duration breakdowns are marked private. No live basket reconstruction was therefore possible. Maximum live basket depth, depth distribution, recovery spacing, lot progression inside live baskets, synchronised closing timestamps, simultaneous opposite-direction exposure and entry-hour distribution could not be measured. Take-profit modifications, the defining mechanism of the tested recovery logic, are not observable in live-monitoring data at all and cannot be compared.

Myfxbook reports a Z-score of −33.28 at 99.99% probability. That is consistent with non-random clustering or dependence in the win/loss sequence; it does not identify the mechanism responsible for that dependence.

Two aggregate differences stand out: the live account trades about 60 positions a month against 80.7 in the post-publication segment, and its average loss is 9.10 pips against 13.91. The vendor states that the account is continuously monitored, regularly optimised, and that trading is paused around news and geopolitical events. The lower trade frequency and different loss profile are consistent with that description, but the aggregate statistics do not establish that these interventions caused the difference — different parameters, broker conditions, version changes, execution, market sample and account size are alternative explanations the available data does not separate.

The vendor-managed operating context

The SignalStart listing describes the account as automated trading by Opal EA “continuously monitored and regularly optimized by the developer”, and states that the operator tracks news and geopolitical events and pauses trading during strong market fluctuations, citing the US election. The Relevant Trading site states that its robots are continuously monitored and that manual intervention occurs from time to time.

These are vendor operating descriptions rather than independently verified facts, and they do not contradict “fully automated” — automated execution and human oversight coexist routinely. They do define what the live record can support. It reflects Opal operating inside a vendor-managed process. It is not evidence of what an untouched default configuration would produce, and the live Profit Factor of 1.82 against the test’s 1.34 cannot be attributed to the EA alone.

Three things stay separate for that reason: the EA’s built-in news filter, whose historical operation is not validated by this backtest; the operator’s stated manual pausing; and the live account result, which reflects both plus broker and sample differences.

The previously associated MQL5 signal

MQL5 signal 2116681, named Opal, is a separate vendor-associated account that the seller’s MQL5 profile linked on 6 September 2025 when announcing the Opal MT5 version, under “Live account”. It is therefore evidence for the Opal MT5 operating record, not validation of the MT4 configuration tested here. No statistics from that signal are used in the MT4 backtest-versus-live comparison.

Non-price deductions and execution sensitivity

An expected payoff of $0.08 per position against a 1.5-pip target leaves a narrow cost margin, and the accounting deserves precision.

Executed-price P/L across the test was $1,123.46 against booked P/L of $574.71. About $548.75 of non-price deductions are inferred from that difference. Roughly $517.58 reconciles to commission under the tested account schedule — 73.94 lots at $7.00 per round-turn lot — with the remaining $31.17 consistent with a net non-commission adjustment (swap/carry) and rounding. Spread is already embedded in the executed entry and exit prices and is not separately isolated by this calculation. Commission is therefore the largest separately reconstructed non-price deduction, not necessarily the largest cost in total.

On a typical winning single-position trade, the strategy captures $0.15 of price movement and books $0.08.

A static sensitivity overlay — holding the trade path fixed, assuming identical entries and exits, and ignoring any interaction with the spread filter — indicates that a schedule of $8 per round-turn lot would reduce the six-year result by roughly $74, and $10 per lot by roughly $222. This is not a rerun. In an actual rerun with different costs or a wider spread, fills, basket formation, take-profit distances and timing would all change, and the spread filter could suppress entries that occurred in this test. Whether a commission-free account with a wider spread would have produced a profitable result cannot be determined without running it.

Two further costs behave differently. The residual non-commission adjustment applies mainly to baskets held across days and, as shown above, changed the sign of every deep buy basket in this sample. Slippage was not simulated; on 1.5-pip targets taken at the hour, a consistent fraction of a pip would be material, and quantifying it requires a separate test.

Because commission is the largest separately reconstructed deduction here, a cashback arrangement has arithmetic relevance to this specific cost line: it can return part of eligible broker commission. It does not reduce spread, remove swap, prevent a basket from expanding, reduce floating exposure or alter the entry logic.

Vendor claims and independent evidence

Two entries need expanding. “AI-driven calculations” cannot be confirmed or refuted from a trading record, since nothing in an order sequence distinguishes a learned scoring function from conditional rules; it is recorded as not independently verifiable. And the $1,000 test does not validate the vendor’s $100 minimum-deposit statement. The observed $306.74 equity drawdown alone exceeded $100 by more than three times, so a separate $100-account test specifying leverage, margin requirement, lot sizing and stop-out conditions would be required to establish what happens at that account size. Scaling the $1,000 result down is not a substitute.

What still needs validation

What a trader would need to validate before live use

The risk profile measured here imposes specific operational requirements.

• Account sizing. The vendor’s $100 minimum is not supported by any test in this review. Establishing a workable balance requires a test at the intended size with specified leverage and stop-out level.
• Cost verification. With $548.75 of inferred non-price deductions against $574.95 booked, the broker’s commission schedule and typical EURUSD spread need measuring on the intended account rather than assumed.
• Floating exposure tolerance. A basket reached eight positions and 0.21 lots, was held 21.7 days across a New Year, and carried an equity drawdown consistent with $306.74 on a balance of about $1,410.
• Absence of a broker-side stop. No standard Stop Loss was recorded on any tested order and the equity-protection switch was disabled in the tested configuration. Anyone requiring a hard per-order stop would need to establish whether one can be configured.
• Direction-dependent holding cost. Deep long baskets in this sample were pushed negative by a residual non-commission adjustment consistent with swap, so applicable swap rates on the intended broker matter for multi-day exposure.
• Operating assumptions. The live monitoring reflects a vendor-managed process including stated pauses. Running the EA unattended is a different proposition from the one the live record documents.

What would be tested next

1. Rerun the identical configuration on $100, with leverage, margin and stop-out conditions specified. Measure whether every position executes, margin level at depth, whether a stop-out occurs, and maximum survivable depth.

2. Measure floating equity directly. The drawdown attribution above is a consistency argument, not a measurement. A run logging equity per tick would give the actual distribution of open exposure.

3. Reconstruct a live basket sample. Whether the deployed account trades like the tested system at basket level remains open because public data does not expose it. A full live history export would resolve depth distribution, spacing, lot progression and synchronised exits.

4. Cost and slippage stress as actual reruns, not overlays, across commission schedules and spread assumptions, including spread-filter interaction.

5. Parameter sensitivity on Scale, Step1, Step2, Step, GridTP, MaxOrders, TakeProfit and the time window.

6. News and year-end behaviour with a controlled event feed, to establish whether features whose historical operation was not validated in this backtest operate as documented.

7. Broker comparison between IC Markets and Vantage or another RAW environment, since the live account runs on a different venue.

Conclusion

The historical test was profitable: $574.95 over six years on EURUSD H1, every calendar-year segment in the sample positive (the complete years 2021–2025, with 2020 and 2026 partial) across 7,033 positions, both directions profitable, and a segment after the publication date that stayed profitable at a similar Profit Factor and win rate. Two advertised features were testable against the trade record and behaved consistently with their descriptions — the time filter, which restricted every one of 6,009 new cycle entries to the configured window, and whole-figure avoidance for initial entries. The aggregate live statistics show a similar win rate, holding-time and payoff profile; in the ten accessible history rows, position size, visible Stop Loss and entry timing were also broadly compatible with the tested configuration, though that small sample cannot establish those characteristics across the full live history.

The remaining uncertainty is concentrated in four places. The recovery tail: 20 baskets reached five positions or more, they were net negative, and nothing in the record establishes what happens beyond the observed maximum depth. Execution sensitivity: inferred non-price deductions were comparable in size to the booked result, and no rerun under alternative cost assumptions has been performed. Operating differences: the live account runs on a different broker inside a vendor-managed process with stated pauses and optimisation, so it does not document an untouched default configuration. And the absence of an independently reconstructed live basket sample, which leaves the central mechanical comparison unresolved.

Forward testing adds evidence about current-market behaviour and execution that a historical backtest cannot provide, and the two evidence layers are complementary — see why forward testing an Expert Advisor is important. Other multi-position systems tested to the same standard are collected in the Forex EA Database.

Test archive

The complete MT4 report, full order history and derived cycle reconstruction for this test are archived at

Сообщение Opal EA Review: MT4 Backtest, Recovery Grid & Live Results появились сначала на EA ForexLab.

Subscribe to our channel, here you will find the best 👇

Understanding why forward testing an expert advisor is important is one of the most useful mindset shifts in algorithmic trading. Many traders follow the same path. First comes an idea. Then comes a beautiful backtest. After that, there is optimization, a strong parameter set, an impressive equity curve, and the feeling that the system is almost ready for a live account. That is exactly where a dangerous illusion often appears: the belief that the strategy tester has already provided enough proof.

In practice, it has not.

Why forward testing an expert advisor is important becomes clear the moment you realize how much distance exists between a strong historical result and real trading. In a backtest, the strategy lives inside a model. In live trading, it meets execution delays, missed ticks, platform differences, changing market regimes, broker infrastructure, and all the details that a strategy tester either simplifies or does not reproduce fully.

From a mature trading perspective, forward testing is not a decorative step after optimization and it is not a nice report for self-confidence. It is a test of whether the strategy can survive after leaving the strategy tester and entering a live environment.

Why a Backtest Alone Is Not Enough

The main problem with a good backtest is that it creates confidence too early. This is especially dangerous when an expert advisor shows a smooth curve, low drawdown, and steady growth. Even a strong historical result still does not answer the most important question: will the logic of the system survive outside the training period?

That is where forward testing becomes essential.

It helps answer whether the strategy has transferability. In other words, does it work not only on the section of history used for parameter selection, but also on the next section that was not used in development? If a system looks convincing on historical data and then quickly falls apart on the next segment, that is a strong sign of overfitting, weak structure, or a strategy that depends too heavily on one market regime.

In practical terms, forward testing does not confirm a strategy once and for all. It does something more modest but much more useful: it shows whether there are enough reasons to keep trusting the system after it leaves the comfortable part of history.

A beautiful backtest usually answers only one question: what would have happened inside the model?
Forward testing begins to answer a much harder question: what happens when the model ends?

What Forward Testing Actually Verifies

Many beginners think forward testing is only about one thing: whether the strategy makes money or loses money on the next period. In reality, it shows far more than that.

First, it shows whether the expert advisor was too heavily fitted to historical data. This is the most basic level. If the system cannot hold itself together on the next period after optimization, there is very little reason to trust that parameter set.

Second, forward testing shows how well the strategy’s behavior transfers from the strategy tester into the live environment. This is where things become much more interesting, because in real trading it is not only the market idea that matters. The engineering implementation matters too.

Third, forward testing shows how dependent the strategy is on specific infrastructure: the broker, the server, the symbol, the platform, tick quality, execution mode, VPS conditions, and even trading hours. This is especially important in cases where it first looks like “the market changed,” but the real issue is the environment where the strategy was deployed.

That is why forward testing is not only a profitability check. It is also a test of whether the research process itself has misled the trader.

Forward Testing Shows Tradability, Not Just Profitability

This is one of the most important practical conclusions and deserves to be stated directly.

Many traders look at forward testing as a mini-exam in profitability: did the expert advisor make money on the next segment or not? A more mature view is different.

Forward testing shows tradability better than it proves theoretical truth.

That is a subtle but critical difference. Forward testing does not mathematically prove that a strategy has a permanent edge. What it shows is something more practical: can this strategy actually be traded in a live environment without its logic being destroyed?

A strategy can look elegant in the tester and logical on paper, but still be poorly tradable under real conditions. If its advantages disappear because of spread, rejects, delays, tick structure, or platform behavior, then in practice it is not a workable trading system, even if the model looked convincing.

That is exactly why forward testing an expert advisor is important before live trading. It is not only about future PnL. It is about whether the strategy can exist as a real operating system in the market rather than as a good-looking idea in a tester report.

Why Matching Live EA Behavior to the Strategy Tester Matters

This is probably the strongest engineering principle behind the entire topic.

Inside the strategy tester, everything looks clean. The required ticks arrive, the logic is processed, the reversal happens where it should happen, and exits occur under the conditions the developer intended. On a live account, that correspondence can break.

And that leads to an uncomfortable but important conclusion: if the live expert advisor does not behave the same way as the tester model, then comparing backtest results to live results becomes only partially meaningful.

At that point, the issue is no longer just whether the strategy is good or bad. The problem is that one object was researched, but a somewhat different object is actually trading.

This is another reason why forward testing an expert advisor is important. It helps verify the quality of the transition from the tester environment to the live trading environment. It reveals whether the strategy’s logic is still intact after deployment.

If the live system begins reacting differently to ticks, entering differently, exiting differently, handling reversals differently, or behaving differently in critical moments, then the question becomes wider: can you still trust the tester results as a valid reference for this implementation?

Missed Ticks, Reversals, and Trading Path Divergence

One of the most underestimated technical problems is that the real trading path can start diverging from the modeled path not because the market changed, but because of small technical mismatches that accumulate over time.

1. A tick did not arrive.
2. A reversal did not happen the same way as in the model.
3. An action was not triggered in the same event sequence as in the tester.
4. An order was not executed at a critical moment.
5. An exit happened at a different price or in a different state of the event flow.

At the level of one trade, this may look minor. But for sensitive systems, these small deviations quickly turn into a completely different trading path. At that point, the live EA is no longer the same system that was tested.

This is why forward testing should never be reduced to a simple “profit or loss” check. It is also needed to answer a deeper question: does the real trading path still match the logic that was verified in the tester?

If it does not, the issue may be far more serious than it first appears. The strategy may not have “stopped working.” Its live implementation may simply no longer match the model that was researched.

Why Execution Quality Matters in Forward Testing

This is one of the most practical conclusions in the entire discussion.

Very often, when a strategy performs worse during forward testing, the trader assumes the market simply changed. Sometimes that is true. But in many cases, the real reason lies elsewhere: in execution quality.

Once the system enters live conditions, many things can destroy the math of an otherwise valid idea: rejects, partial fills, differences between virtual and real trading logic, order-processing delays, changing spread, rollover effects, and differences in how market and limit orders are handled.

For execution-sensitive systems, these are not secondary details. They are the core of the problem. A beautiful backtest may become almost useless if live orders are not executed in the way the model assumed. That is why forward testing is also a test of whether the strategy is executable in practice, not only whether it is profitable in theory.

This is a critical shift in understanding. Forward testing an expert advisor is important not simply because it shows what the system may earn on the next market segment. It is important because it reveals whether execution destroys the strategy’s internal mechanism.

Why Real Ticks Are Critical

If a strategy is sensitive to tick-level behavior, then rough testing modes can create dangerous illusions. This becomes especially obvious in systems that look excellent on generated ticks but deteriorate sharply on real ticks.

A good example of a vendor test in the MetaTrader 4 terminal with 90% quality and low quality tick history:

Test MT4 Tick Data Suite 99% Real Spread with the highest quality tick history from broker Darwinex:

This affects forward testing as well. The value of forward testing depends heavily on the quality of all prior research. If the strategy was originally studied in an overly rough model, then the forward stage already stands on a weak foundation.

That is why good forward testing cannot be separated from the subject of real ticks. It should be the continuation of a research process that was built as close to real trading conditions as possible from the start.

The practical meaning is simple: the weaker the model before forward testing, the less confidence you should have in the forward test itself. If the strategy received a false sense of credibility in the tester, then the next validation stage is already contaminated by that distortion.

Why One Good Out-of-Sample Period Can Be Misleading

Traders naturally enjoy finding confirmation. The problem is that a single successful out-of-sample period can very easily become a tool of self-deception.

A favorable segment may have been chosen by accident. The process may have been structured in a way that weak parameter sets were filtered out while attractive ones were kept. The OOS period may simply have been a segment where almost any reasonable system would have looked profitable.

That is exactly why one forward test should never be treated as a seal of quality. It may be a useful signal, but by itself it does not protect against selection bias, luck, or subtle forms of looking into the future.

A serious research process begins when a trader stops searching for confirmation and starts searching for reasons to doubt the result.

Why Walk-Forward Testing Is More Honest Than a Single Forward Test

This is where walk-forward testing becomes especially valuable.

A one-time optimization plus one forward period can look promising, but walk-forward testing usually gives a picture that is much closer to real operation. Yes, it often looks less beautiful. But it is also more honest.

Walk-forward testing is useful because it forces the strategy to go through a repeated cycle: optimize, test on the next segment, re-optimize, test again. This makes it much harder to hide behind one lucky period. It gives a better picture of how the strategy might behave in real life, where the market keeps changing and the trader does not get to live inside one attractive historical fragment.

Put simply, a single forward test answers: “maybe?”
Walk-forward testing gets much closer to: “how is this likely to behave in actual use?”

That is why walk-forward testing usually gives a less glossy but more truthful view of future behavior.

Common Forward Testing Mistakes

There are several mistakes traders make again and again.

The first is treating forward testing as a guarantee. It is not a guarantee. It does not promise future profit. At best, it reduces uncertainty if done honestly.

The second is drawing conclusions from one attractive out-of-sample period. One successful segment can still be nothing more than a favorable segment.

The third is ignoring execution quality. A forward test without execution analysis is only a partial forward test.

The fourth is building conclusions on top of a weak testing model. If the system was studied in an unrealistic framework before forward testing, then the final validation stage is already compromised.

The fifth is leaking future information through the selection process. Rejecting inconvenient candidates after the fact often looks like reasonable filtering, but in reality it becomes disguised overfitting.

The sixth is reducing the entire analysis to profit and drawdown while ignoring whether the live trading path still matches the logic tested in the strategy tester.

What Forward Testing Should Prove Before Live Deployment

Before a strategy is trusted with real money, forward testing should help answer several practical questions.

• Does the expert advisor remain stable outside the training sample?
• Does the live behavior still match the logic validated in the tester?
• Does execution degrade the system’s edge?
• Do broker and platform conditions distort the strategy?
• Is the system too sensitive to infrastructure such as symbol, server, VPS, or trading hours?
• Does the research process still look trustworthy after this additional layer of validation?

That is the correct role of forward testing. It does not prove perfection. It proves whether the system deserves the right to move one step closer to live deployment.

Conclusion

Forward testing an expert advisor is important not because it provides a comforting screenshot after optimization. Its real purpose is to verify whether the logic discovered in the strategy tester can survive in an environment where everything becomes harsher: real ticks, real execution, real platform limitations, and real implementation errors.

That is why forward testing shows more than most traders think. It does not only show profit or loss. It shows robustness, overfitting risk, tradability, execution sensitivity, engineering correctness, infrastructure sensitivity, and the level of trust you can place in the entire research chain.

A strong algorithmic trader is not the one who finds the most beautiful backtests. A strong algorithmic trader is the one who understands that between a good backtest and a real trading system, there must be an honest forward test.

You can see a catalog of advisors that we tested with real spreads on high-quality tick history on this page.

Subscribe to our channel, here you will find the best 👇

Сообщение Why Forward Testing an Expert Advisor Is Important Before Live Trading появились сначала на EA ForexLab.

The more important risk number is equity drawdown. Balance drawdown was 2.37%, while equity drawdown reached 13.83%. Reconstructing the grid cycles from the deal history shows why the gap is so large: some baskets added several same-direction positions and remained open for weeks while carrying floating losses.

The MQL5 description says the grid is designed so that every trading cycle concludes with a win. In this test, all 587 reconstructed cycles had positive realised P/L before commission and swap, but 5 finished negative after account costs. That difference between trade P/L and final account contribution is central to the review.

Key findings

• It is a grid. The order comments run QB (Step: 1), (Step: 2), (Step: 3) and so on. I reconstructed 587 complete trading cycles, with additional same-direction positions appearing as price moved against the first entry.
• Balance drawdown alone understates floating equity risk. Maximum balance drawdown was 2.37% ($310.76); maximum equity drawdown was 13.83% ($1,813.13) — roughly 5.8 times larger. The open-equity path was much rougher than the balance curve suggests.
• Most cycles closed on the first entry. About 62% of cycles closed without requiring a Step 2 entry. The remaining 38% required at least one additional position.
• “Every cycle wins” depends on how a win is counted. All 587 cycles were positive before commission and swap; after account costs, 582 remained positive and 5 were negative. The MQL5 description does not define whether a win is measured before or after costs.
• Costs are material. Commission (−$729.92) and swap (−$293.38) removed more than $1,000 from the pre-cost trade result over the test.
• Broker environment matters. This test ran on RannForex-Server; the MQL5 page recommends IC Markets / IC Trading. The commission, swap and BTCUSD conditions here describe the tested environment and may not reproduce on another broker.
• The deepest observed basket reached Step 7. InpGridMaxTrades was configured at 10, but the historical test itself only establishes the observed depth.

What tested

This is an independent EA ForexLab test of Quantum Bitcoin EA for MetaTrader 5, listed on the MQL5 Market by Bogdan Ion Puscasu. The current product page describes it as a BTCUSD H1 trend-following strategy with grid-based position management and says the grid is designed so that every trading cycle concludes with a win. It currently recommends a $1,000 minimum deposit, 1:500 leverage, a hedge account and a low-spread ECN/Raw/Razor account.

I focused on what the headline report does not show: grid depth, cycle structure, trading costs, holding time and floating equity exposure. The methodology behind the test is described in the EA ForexLab testing methodology for trading algorithms.

Test setup

Every value below comes directly from the Strategy Tester report. The test ran on RannForex-Server (build 5833) with the configuration string Quantum Bitcoin EA v3.2 (16/02/2026). The Strategy Tester reports 100% real-tick history across 31,900 bars and approximately 140.9 million ticks.

One input worth flagging: InpSlippage=10 is an EA/test input. It is not evidence that realistic historical market slippage was modelled, and I do not treat it as such.

I ran this test on RannForex-Server, while the MQL5 page currently recommends IC Markets or IC Trading. The commission, swap and BTCUSD contract conditions analysed here therefore describe the tested RannForex environment and should not be assumed to reproduce unchanged on another broker.

MT5 backtest results

The headline Strategy Tester statistics are strong.

Net profit was $3,934.46 on a $10,000 start, with a Profit Factor of 2.69 and 757 of 975 trades closing in profit. Both directions were profitable: shorts won 78.04% of 542 trades and longs won 77.14% of 433. The largest single winner was $223.87 and the largest single loser −$78.91, so no individual trade dominated the result.

I include both drawdown figures because they describe different parts of the risk. For a multi-position grid, the distinction matters.

Why I care more about equity drawdown than balance drawdown

Balance reflects realised account results, while equity also reflects open floating P/L. A grid can amplify the difference because several positions may remain open simultaneously while a basket moves against the strategy. Here the two curves diverged sharply.

Maximum balance drawdown was 2.37%. Maximum equity drawdown was 13.83%, a ratio of about 5.8 to 1. The closed balance therefore looks much smoother than the equity path: the open-position risk reached a level the balance curve does not capture.

The two drawdown figures measure different things. For this grid, equity drawdown is the more informative historical risk measure because it includes floating P/L from open baskets. The observed 13.83% should not be treated as a future maximum.

How Quantum Bitcoin EA actually builds a grid

I reconstructed the grid structure from the order comments and the transaction history. Each observed cycle starts with a Step 1 entry, and additional same-direction Step entries appeared as price moved against the existing basket. The inputs include grid-distance and distance-multiplier parameters, but the deal history does not expose the exact internal trigger. Positions within each reconstructed cycle closed together at a common exit price; the report shows the observed behaviour, not the exact internal exit rule.

The position-level win rate and the cycle-level result measure different things. A multi-entry basket can contain both winning and losing positions and still finish positive as a group. That is how a 77.64% position win rate can coexist with 587 reconstructed cycles that were all positive before commission and swap. In total, 218 individual positions closed at a loss.

How often the grid needed additional entries

Grid frequency and depth are two important risk diagnostics, so I measured both. The table below is the deepest Step each cycle reached before closing.

About 62% of cycles closed without a second entry. Most of the rest reached Step 2 or 3. Only 6 cycles reached Step 6 or deeper, and exactly one reached Step 7. The reconstructed cycle chronology showed no overlapping baskets. Maximum simultaneous exposure was seven positions, reached by the single Step 7 cycle.

Most cycles ended after Step 1, but 37.82% required at least one additional entry. Deeper baskets deserve particular attention because they carry more simultaneous exposure and can remain open long enough for financing costs and floating drawdown to become material.

Does every Quantum Bitcoin trading cycle really win?

The vendor’s cycle claim can be checked directly against the reconstructed ledger. I calculated each cycle two ways: realised trade P/L before commission and swap, then final account contribution after those costs.

All 587 reconstructed cycles had positive realised P/L before commission and swap. After account costs, 582 remained positive and 5 were negative. The MQL5 description does not define whether a “win” is measured before or after trading costs.

Put concretely: if a winning cycle is defined before commission and swap, all 587 qualified in this historical test. If it is defined by final account contribution, 5 of 587 did not.

Commission and swap change the answer

Here is the full reconciliation, straight from the deal ledger. Aggregate trade profit and loss before costs was +$4,957.76. Commission took −$729.92. Swap took −$293.38. Net result: +$3,934.46, which matches the report to the cent. The $4,957.76 figure is the net sum of the MT5 Profit column before commission and swap. It is not the Strategy Tester Gross Profit metric of $6,263.46, which sums profitable trades only.

Commission accumulated to −$729.92 across the 1,950 deals in this test. Swap added −$293.38. Swap can accumulate when positions remain open across the broker’s financing periods, which makes prolonged baskets especially sensitive to holding costs. Trading costs turned five raw-positive cycles into net-negative cycles: swap was the decisive cost in four of the five, while one cycle was already negative after commission.

This test also shows why trading costs matter for the strategy. On an eligible broker and account, a forex rebate can reduce part of eligible commission. It does not reduce swap or grid exposure, and I do not assume the RannForex commission in this test is rebate-eligible.

The deepest grid sequence in the test

The deepest sequence was a Step 7 sell basket. Its first position opened on 23 June 2025 and the group closed on 26 August 2025 — a basket span of about 63 days from first entry to final closure.

The entries were sized 0.01, 0.01, 0.01, 0.02, 0.02, 0.02 and 0.03 lots as the grid deepened, opened at prices climbing from roughly 102,500 up to 117,600 as BTC ran against the position. By the time all seven positions closed together near 109,600, BTC had retraced substantially from the highest grid-entry area. The exact internal exit condition is not exposed by the report.

The basket closed with +$53.51 in trade P/L before costs but finished at −$37.25 after costs. The prolonged holding period accumulated $77.57 in swap, and commission added another $13.19. Basket duration and position duration should also be kept separate: this basket spanned about 63 days from first entry to final closure, while the longest individual position anywhere in the test lasted 1083 hours 33 minutes, about 45 days.

Position sizing and the 1.2× multiplier

The inputs set InpInitialLots=0.01, InpLotsMultiplier=1.2 and InpLotsMax=1.2. Realised positions ranged from 0.01 to 0.05 lot; most were 0.01 or 0.02. InpLotsMax was configured at 1.2, while the largest realised position in this test was 0.05 lot. I do not infer the exact internal meaning of InpLotsMax from the parameter name alone.

I would classify the observed structure as a progressive averaging grid rather than a classic 2× martingale. In this backtest, the main exposure increase came from accumulating same-direction positions rather than a doubling lot sequence. That distinction does not remove grid risk: exposure still rises as additional positions are added.

There is also an anomaly in the sizing worth recording. The configured InpInitialLots value is 0.01, yet every Step 1 entry in 2022 opened at 0.02 lot. From April 2023 onward, Step 1 used 0.01. The cause is not established by the test, so I do not assign one. What is confirmed is a mismatch between the configured initial lot and the realised early-period order size.

Long vs short

Both directions were profitable. The report shows 542 short trades with a 78.04% win rate and 433 long trades with a 77.14% win rate. Reconstructing the account contribution by direction gives about +$2,339.78 from sell cycles and +$1,594.68 from buy cycles after costs.

The larger sell contribution does not establish better short-side behaviour. Sell cycles were more numerous over the test, while the long and short win rates were almost identical. Both directions remained net positive after costs, so the total result was not produced by one side offsetting a losing opposite side.

Trade duration

This is not a scalper. The report puts the minimum position holding time at 56 seconds, the average at about 24 hours 36 minutes, and the maximum at 1083 hours 33 minutes — roughly 45 days.

The distribution is concentrated in the hours but has a long tail. A position held for days or weeks remains exposed to BTC price movement for longer, and financing costs can accumulate when positions remain open across the broker’s financing periods. Long duration alone does not determine whether a position finishes profitable or losing. It matters here because deeper baskets can combine prolonged market exposure, accumulated swap and larger floating losses.

Trading hours and weekend behaviour

The inputs set InpStartTime=08:00 and InpEndTime=22:30 in server time, with InpTradingFriday=false. Because BTCUSD trades through the weekend in this test environment, I reconstructed activity by hour and weekday to compare the observed orders with those settings.

All 587 Step 1 entries occurred between 08:00 and 22:00 server time, while Step 2 and deeper entries also appeared outside that window. This pattern is consistent with the configured time window applying to new-cycle initiation while existing baskets continue to be managed.

Friday is the clearest case. No Step 1 entry occurred on a Friday, while additions and exits did, and new cycles opened on Saturdays and Sundays. This is consistent with InpTradingFriday=false affecting new-cycle initiation rather than all basket activity.

Performance by year

I grouped realised deal P/L, commission and swap by the calendar year in which each account transaction was booked.

Every displayed period contributed positively, but the profit was unevenly distributed. 2024 and 2025 contributed most; 2022, 2023 and the partial 2026 were much smaller. There was also a long inactive interval: no new Step 1 cycle opened between 22 September 2022 and 12 April 2023 — roughly 202 days. Five positive calendar segments are useful evidence, but they are not proof of robustness across market regimes.

Vendor claims vs test evidence

The table below compares the current MQL5 positioning with what the independent test establishes.

What I would test next

The historical test answers several structural questions, but live transferability still needs separate validation. I would prioritise the following tests.

• Forward test on the intended broker and account. This is the next direct check of current execution conditions: spread, charged commission and swap, slippage behaviour and grid-depth frequency. If the goal is to measure actual live execution rather than demo execution, the measurement needs to come from a live account. The EA ForexLab guide explains why forward testing an Expert Advisor matters.
• Capital sensitivity: $1,000 vs $10,000. The test used $10,000 while the MQL5 page states a $1,000 minimum. If the same absolute position-sizing behaviour were used on $1,000, percentage exposure and margin pressure would be materially different. The $1,000 case therefore needs a dedicated backtest rather than an extrapolation from this one.
• Adverse-move stress. The deepest basket in this test reached Step 7. A sustained adverse BTC move is the most relevant scenario for testing whether the strategy can progress beyond the historical maximum observed here.
• Broker comparison. BTC contract specifications, financing costs and spreads can differ materially between brokers. That matters for a grid that can keep positions open for weeks.
• Parameter sensitivity. Grid distance, the distance multiplier, the lot multiplier, the InpGridMaxTrades setting and the trading-hours window all affect the observed exposure pattern and are worth testing systematically.

Who this EA may suit — and who it may not

This is a description, not advice, and nothing here is a recommendation to buy or to trade.

The profile is more relevant to experienced grid users who monitor floating equity, understand prolonged BTC exposure and are prepared to validate broker-specific costs. It is a poor fit for anyone judging the system from win rate or balance drawdown alone, or for anyone uncomfortable with several same-direction positions being held at once. This $10,000 test should not be used as validation of the vendor’s $1,000 minimum.

Final verdict

The historical result is strong under the tested conditions: $3,934.46 net profit, Profit Factor 2.69, a 77.64% trade win rate, positive contribution from both directions and positive account contribution in every displayed calendar period. The deepest observed basket reached Step 7, while InpGridMaxTrades was configured at 10.

The risk profile is still a grid risk profile. Equity drawdown was almost six times the balance figure, and some baskets remained open for weeks. Trading costs turned five raw-positive cycles into net-negative cycles, with swap the decisive cost in four of them. A longer adverse BTC move remains an important untested stress scenario, particularly because the deepest historical basket already reached seven positions and lasted about 63 days. The $10,000 backtest also does not establish the risk profile of the vendor’s $1,000 minimum.

My conclusion is straightforward: Quantum Bitcoin EA should be evaluated by its equity exposure, grid depth and cost structure rather than balance drawdown alone. The backtest documents a positive historical result, but it does not show how the same grid will behave under current live broker conditions. The next useful step is a forward test on the intended broker and account.

The independent test conditions and configuration are catalogued in the Forex EA Database. For a structurally comparable grid from the same product family, the Quantum Athena EA review covers the gold-side sibling and the same balance-versus-equity question.

Test archive

This review is based on a MetaTrader 5 real-tick Strategy Tester run of Quantum Bitcoin EA v3.2 on BTCUSD H1 from 2022 to 2026. I analysed the complete deal history and reconstructed the grid cycles, trading costs, position depth and timing from those transactions. Vendor positioning is referenced to the current official MQL5 product page.

This article documents an independent backtest for research purposes. It is not financial advice, not a recommendation to buy, and not a prediction of future results. Automated trading on a leveraged, volatile instrument like Bitcoin can lose money.

Сообщение Quantum Bitcoin EA Review: BTCUSD MT5 Backtest & Grid Risk Analysis появились сначала на EA ForexLab.

🔍 From subscriber‼️

🤖 EA name: Pac Man
📦 Version: 4.59
💻 Platform: MT4 (1470)
🛠Vendor/Source: –
📈 Strategy: Order grid
⏰ Timeframe: m15
🌍 Currency pairs: AUDCAD, AUDNZD, NZDCAD
🌓 Trading time: Around the clock
⚠️ Attention: Recommended best VPS, BROker
📊 Monitoring found: –

⏳ Test period: 2020.01.01 – 2026.04.19
🏛 Tick Data Provider: Darwinex (TDSv2)
🧭 GMT: +2; DST: US
Real spread: ✅
Slippage: ❌

📌 All EAs tests EA ForexLab

In order to download an adviser with tests, go to our telegram channel 👇

The retail market is full robots that promise precision, stability, and “smart” automation, yet most reviews still make the same mistake: they compare headline profit without examining how that profit was produced. You can learn about our approach to testing expert advisors on tick history with a real spread on the relevant page – Our principles.

Pac Man EA Review: Read This Before You Trade

This Pac Man EA review is different from the ones you see on seller sites. Those show a clean rising chart and a buy button. This one shows the full picture.

We tested the Pac Man EA on two currency pairs. We read every trade in both backtest files. We checked every setting. And We found something the marketing never mentions.

On one pair, AUDCAD, the robot looked great. It grew a $1,000 account to over $2,000 across six years. On the second pair, AUDNZD, the same robot crashed a $1,000 account down to just $88. That is a 91% loss.

Same robot. Same settings. Two completely different results. Let me explain why this happened, and what it means for your money.

What Is the Pac Man EA?

The Pac Man EA is an automated trading robot for MetaTrader 4. The version tested here is V4.59. It trades on the M15 (15-minute) timeframe. Its trade comment is “Waka,” whatever that means.

The robot is built to trade currency pairs like AUDCAD, AUDNZD, and NZDCAD. These are known as “mean-reversion” pairs. They tend to move in ranges and return to an average price. Many grid robots target these pairs for that reason.

The marketing presents Pac Man as a smart, multi-pair system. It promises steady growth and built-in protection features. But the real story lives in the trade data, not the sales page. So let me show you the data.

The backtest used Tick Data Suite with real tick history from the Darwinex broker. It ran on an IC Markets terminal. The modeling quality was 99.90%. This is a strong, honest testing method. The problem is not the test. The problem is what the robot actually does.

Pac Man EA Review: The Two-Pair Tale

The clearest way to understand this robot is to compare its two tests side by side.

Look at the drawdown numbers. On AUDCAD, the worst drop was 7.55%. That is manageable. On AUDNZD, the worst drop was 93.90%. That is a near-total wipeout.

The win rates are almost the same. Both pairs won about 75% of trades. So how can one make money and the other destroy the account? The answer is the grid. And that is the heart of this Pac Man EA review.

The Hidden Engine: A Martingale Grid

Here is the most important thing to understand. The Pac Man EA is a grid robot that uses martingale-style position sizing. The marketing does not say this clearly. The trade data proves it beyond doubt.

Let me explain what that means in simple terms.

A grid robot opens more trades when the price moves against it. The idea is to lower the average entry price. When the price finally turns around, all the trades close in profit together.

The “martingale” part is about trade size. With each new grid level, the robot opens a bigger position. I traced the exact lot sizes in the AUDNZD test. Here is the sequence the robot used:

0.01 → 0.02 → 0.04 → 0.08 → 0.12 → 0.20 → 0.32 → 0.52

Each step is about 1.5 to 2 times larger than the last. By the eighth level, the robot was trading 0.52 lots. That is 52 times the starting size. The settings confirm this with a TradeMultiplier value of 1.3 and higher.

This design has a clear pattern. It wins small and often. Then, once in a while, it loses big. Very big. The AUDNZD test is what “big” looks like.

How AUDNZD Died: A Step-by-Step Breakdown

The AUDNZD account did not lose money slowly. It collapsed in a single grid event. Here is what the data shows.

In April 2024, the price moved hard against the robot’s positions. The robot did what grid robots do. It kept opening larger and larger trades to defend its losing basket.

At the peak, the robot had 11 trades open at the same time. The largest was 0.52 lots, on an account that started with just $1,000. The combined loss grew faster than the account could handle.

The single worst trade lost $389.69. To put that in context, the average winning trade on this pair made only a few dollars. One bad trade erased hundreds of good ones.

Then something telling happened. The account balance fell to $88.39. And the robot simply stopped trading. The backtest was supposed to run until April 2026. But AUDNZD made its last trade in April 2024, with nine positions still stuck open. The account was too damaged to continue.

In plain words: the account blew up two years before the test even ended. The smooth equity curve you see in the marketing is the AUDCAD chart. The AUDNZD chart is the one that drops off a cliff.

Why a 77% Win Rate Fooled So Many People

A 77% win rate sounds excellent. Most traders see that number and feel safe. But with a grid martingale, the win rate hides the real risk.

Think about it this way. The robot wins most trades by a small amount. The average AUDCAD win was just $1.47. But when it loses, the loss can be huge, like the $389.69 hit on AUDNZD.

So the math is dangerous. You collect many small wins. Then one grid failure takes back months of profit in a single day. The high win rate does not protect you. It actually masks the danger until it is too late.

Here is the key lesson. A grid robot’s win rate tells you almost nothing about its safety. What matters is the size of the rare loss. And on AUDNZD, that loss was the entire account.

The Trade Timing Confirms the Grid

I looked at how long the robot holds its trades. The Trades by Time chart tells the story clearly.

The biggest cluster of trades closes in the 4-hour window. The next big clusters are at 8 hours, 16 hours, and even 4 days. Some trades stay open for over a week.

This is normal for a grid system. The robot holds losing trades and waits for a reversal. It does not cut losses quickly. It hopes the price comes back. Most of the time, on range-bound pairs, it does. But not always. And “not always” is what kills the account.

A robot that cuts losses fast would never hold a basket for days while it grows. The long hold times are a direct sign of the grid recovery logic at work.

The Win/Loss by Hour Pattern

The Wins/Losses by Hour chart shows another useful detail. The robot trades most actively in the afternoon hours, around 15:00 to 18:00 server time. It also trades through the day with steady activity.

Green bars (wins) clearly outnumber red bars (losses) in most hours. This matches the high win rate. But again, the chart only counts the number of wins and losses. It does not show their size.

A hundred small green wins can be wiped out by a few large red losses during a grid failure. The hourly chart looks healthy. The account result on AUDNZD was not.

Pac Man EA Review: The Settings That Raise Red Flags

The backtest report shows the robot’s full settings. A few of them deserve a close look, because they reveal the true risk level.

First, MaximumOpenLots=100000. This sets a huge ceiling on total open volume. In practice, it means the grid can grow almost without limit. There is no tight cap to stop a runaway basket.

Second, MaximumDrawdown=100. This allows the robot to draw down the entire account before it stops. On AUDNZD, it nearly did exactly that.

Third, the robot is designed to run on multiple pairs at once. The settings list “AUDNZD, AUDCAD, NZDCAD” as a portfolio. This is important. It means AUDNZD is not a random bad choice. It is one of the robot’s own recommended pairs. The disaster happened on a pair the system was built to trade.

Fourth, the news filter and stock market crash filter were available but set in ways that offer limited protection in the test. A grid system needs strong protection. These settings did not prevent the AUDNZD collapse.

A Closer Look at the AUDCAD “Success”

It is fair to give the winning pair a proper review. So let me break down the AUDCAD result year by year. This is the test that the marketing relies on.

On the surface, this looks very consistent. The robot made money every single year. The win rate stayed near 75% to 80% throughout. This is a genuinely steady record on this one pair.

But notice the trend in the later years. The yearly profit was falling. From a peak of $196.69 in 2022, it dropped to $145 in 2024 and $136 in 2025. The number of trades also fell. This could mean the edge is slowly fading, or that recent market conditions suit the robot less well.

Even so, AUDCAD is the good news story. The robot handled this pair without a deep grid failure. The key point is that this success and the AUDNZD disaster come from the exact same code. The difference was the market, not the robot.

The Survivorship Trap

Here is a mistake many traders make. They look at the AUDCAD result, see steady profit, and assume the robot is safe. This is called survivorship bias.

The AUDCAD account survived because its grids stayed shallow. The deepest it went was 0.12 lots, only a few levels in. The price always came back before the grid grew dangerous.

The AUDNZD account did not get that luck. Its grid reached 0.52 lots and kept losing. The price did not come back in time. The account died.

Same robot. Same engine. The only difference was market behavior and luck. If you only see the winning pair, you believe the robot is safe. But the losing pair shows the true risk that was there all along. Both outcomes come from the same design.

Why Grid Robots Target These Pairs

It helps to understand why the Pac Man EA trades AUDCAD, AUDNZD, and NZDCAD. These are not random choices. They are some of the most range-bound pairs in the entire forex market.

The Australian, New Zealand, and Canadian dollars are all linked. Australia and New Zealand are close trading partners with similar economies. Canada and Australia are both commodity-driven nations. As a result, these pairs often move sideways in tight ranges rather than trending for long periods.

Grid robots love this behavior. When a pair stays in a range, every move against the grid tends to reverse. The basket recovers, and the robot books a profit. This is why grid systems can show years of smooth gains on these pairs.

But there is a catch, and AUDNZD showed it perfectly. Even the most range-bound pair can break out. When AUDNZD made a strong directional move in April 2024, the range-trading logic failed. The grid kept adding to a losing position that did not come back in time. The very feature that makes these pairs attractive to grids is also the trap. Ranges work, until they suddenly do not.

This is the core weakness of every grid robot. It assumes the past range behavior will continue. When the market changes character, the assumption breaks, and the losses arrive all at once.

The Spread and Cost Reality

The backtest ran on Darwinex , a broker with very tight spreads. On crosses like AUDCAD and AUDNZD, that matters a great deal.

These pairs naturally have wider spreads than majors like EURUSD. On an average retail broker, the spread on AUDNZD can be two to four times wider than on Darwinex . Each grid trade pays that spread on entry.

Now remember the grid. When the robot opens 11 trades in one basket, it pays the spread 11 times. The wider your broker’s spread, the more each grid event costs you. On a normal broker, the small wins shrink and the deep grids cost even more. The thin edge the robot showed in testing could disappear.

This is why a grid backtest on ultra-tight spreads is a best-case picture. Your real results will likely be worse, not better.

The Real Cost of a Grid Blowup

Let me make the risk concrete with simple numbers.

On AUDNZD, the account fell from $1,000 to $88. To recover from an 88% loss, you do not need an 88% gain. You need a gain of over 700% just to get back to where you started. That is almost impossible for most traders.

This is the cruel math of large drawdowns. Small losses are easy to recover. Huge losses are not. A grid robot that wins for years can still leave you far worse off than when you began, if a single basket fails badly.

And remember, this is a backtest in perfect conditions. Live trading adds wider spreads, slippage, and requotes. A grid that struggled in a clean test will struggle more in the real world.

Pac Man EA Review: Your Pre-Trade Checklist

If you are still thinking about this robot, work through this checklist first. Do it before you risk any money.

First, accept that this is a grid martingale. No matter what the marketing says, the data shows growing lot sizes on losing baskets. Price it as a high-risk system, because it is one.

Second, study the AUDNZD result, not just AUDCAD. The winning pair shows the good days. The losing pair shows what happens on a bad day. Judge the robot by its worst result, not its best.

Third, understand the win rate trap. A 77% win rate does not mean safety. The rare large loss is what matters. One grid failure can erase years of small wins.

Fourth, never use the default risk on a small account. The settings allow the grid to grow toward 100% drawdown. On a $1,000 account, that means losing nearly everything. Use the smallest possible size if you test it at all.

Fifth, demand a long, verified live track record. A backtest is not enough for a grid system. You need to see real results across many years and many market shocks. Ask how the robot behaved in real grid failures.

Sixth, test on a demo for months before going live. Watch how deep the grids go. Watch how long trades stay open. If you see a basket growing fast, you are seeing the AUDNZD pattern begin.

Final Verdict: A Grid Robot With a Hidden Cliff

Let me sum up this Pac Man EA review in plain terms.

The Pac Man EA is a grid trading robot that uses martingale position sizing. On the right pair, in the right conditions, it can produce a long run of steady profit. The AUDCAD test proves it can win for years. That is real, and I will not deny it.

But the AUDNZD test proves something more important. The same robot, with the same settings, crashed a $1,000 account down to $88. It blew up two years before the test ended. Nine trades were still stuck open when it died. And AUDNZD was one of the robot’s own recommended pairs.

This is the nature of grid martingale systems. They look smooth and safe right up until the moment they are not. The smooth equity curve is the trap. The cliff is hidden behind it.

For beginner traders: a rising chart and a high win rate are not proof of safety. The real risk is the rare big loss. With this robot, that loss can be your whole account.

For intermediate traders: the two-pair comparison is your key lesson. Always look for the system’s worst result, not its best. Survivorship bias makes grids look safer than they are.

For experienced traders: you already know the signs. Growing lot sizes, 100% drawdown allowance, no real lot cap, and trades held for days. This is a martingale grid. Without strict risk limits and years of verified live data through real shocks, the AUDCAD profit is not proof of an edge. It is proof of good luck on one pair.

Grid robots can win for a long time. But the question is never whether they will lose. The question is when, and how much. On AUDNZD, the answer was almost everything.

Trade with your eyes open.

Even more advisors with test results are presented in our advisor database.

In order to download an adviser with tests, go to our telegram channel 👇

Сообщение Pac Man Review EA for MT4: A Critical Analysis of the AUDCAD Backtest появились сначала на EA ForexLab.

🔍 From subscriber‼️

🤖 EA name:Quantum Athena
📦 Version: 1.1
💻 Platform: MT5 (5833)
🛠Vendor/Source: MQL5
📈 Strategy: Order grid
⏰ Timeframe: H1
🌍 Currency pairs: XAUUSD
🌓 Trading time: Around the clock
⚠️ Attention: Recommended best VPS, BROker
📊 Monitoring found: MQL5

⏳ Test period: 2023.01.01 – 2026.04.25
🏛 Tick Data Provider: RannForex (MT5)
🧭 GMT: +2; DST: US
Real spread: ✅
Slippage: ❌

📌 All EAs tests EA ForexLab

In order to download an adviser with tests, go to our telegram channel 👇

The retail market is full of gold robots that promise precision, stability, and “smart” automation, yet most reviews still make the same mistake: they compare headline profit without examining how that profit was produced. You can learn about our approach to testing expert advisors on tick history with a real spread on the relevant page – Our principles.

Quantum Athena EA Review: Reading Between the Equity Lines

This Quantum Athena EA review takes a different approach from the dozens of affiliate pages already circulating about this gold robot. Instead of quoting the marketing copy or admiring the smooth equity curve, I extracted all 3,500 individual deals from the MetaTrader 5 backtest report, reconstructed every trade cluster, and identified exactly what is generating the returns.

What I found is a system that is far more dangerous than its statistics suggest. Behind an attractive 81% win rate and a Sharpe ratio of 5.91 sits a grid averaging engine that opened 27 simultaneous positions during a single adverse move in March 2024. The marketing never mentions this. The product description never mentions this. But the trade data makes it unmistakable.

Let’s go through everything — the good, the concealed, and the genuinely concerning.

What Is the Quantum Athena EA?

The Quantum Athena EA is a MetaTrader 5 Expert Advisor developed by Bogdan Ion Puscasu, sold on MQL5.com for $699.99. It was published on April 21, 2026 — making it an extremely new product — and trades exclusively on XAUUSD (Gold vs US Dollar) on the H1 timeframe, though the EA internally manages its own timeframe logic.

The marketing positions it explicitly as “the light version of the legendary Quantum Queen, refined and re-engineered for today’s market conditions.” This lineage is important. The Quantum EA family — including Quantum Queen, Quantum Emperor, Quantum King, and Quantum StarMan — is well known in the retail community, and several of these systems use grid-based position management. Quantum StarMan’s own product page openly states it “utilizes a sophisticated grid system.”

Quantum Athena inherits this DNA. And as the trade data confirms, it is fundamentally a grid system — a fact that materially changes its risk profile compared to how it is marketed.

The backtest analyzed here was run on the RannForex MT5 terminal with 100% real tick history, modeling 493 million ticks across the period January 1, 2023 to April 25, 2026. The history quality is excellent. The period selection, as we will discuss, is not.

Quantum Athena EA Review: The Headline Backtest Numbers

The MetaTrader 5 Strategy Tester reported the following:

At first glance, this looks like an institutional-grade strategy. A Sharpe ratio of 5.91 is exceptional — most professional hedge funds operate with Sharpe ratios between 1 and 2. A recovery factor of 3.63 and profit factor of 2.62 are both solid. The 81% win rate looks reassuring.

But three numbers on this list tell the real story, and they are easy to miss: the gap between balance drawdown (7.25%) and equity drawdown (27.48%), the Z-Score of -3.46, and the 3.3-year period. Each of these is a red flag, and together they reveal the system’s true nature.

The Balance vs. Equity Drawdown Gap: Where the Truth Hides

This is the single most important section of this Quantum Athena EA review, so read it carefully.

The backtest reports two different drawdown figures:

• Balance Drawdown Maximal: $147.11 (7.25%)
• Equity Drawdown Maximal: $557.59 (27.48%)

The balance drawdown measures the decline in realized (closed) account value. The equity drawdown measures the decline in total account value including open, unrealized positions. When these two numbers are close together, a system closes trades cleanly and rarely holds large floating losses. When they diverge dramatically — as here, by nearly 4x — it means the system is holding large baskets of losing positions open and only closing them when they recover.

A 27.48% equity drawdown means that at one point in this backtest, a trader with $1,000 was looking at $557 in combined realized and unrealized losses — more than half their capital tied up in underwater positions waiting to recover. The balance “only” showed a 7.25% decline because the losing positions had not yet been closed and booked.

This is the defining signature of a grid or averaging strategy. The balance curve looks beautiful and smooth because losses are not realized until the grid recovers. The equity curve — the real-time truth of the account — tells a far more violent story.

The 27-Position Grid: Forensic Proof

The product marketing describes “precision” and “intelligent optimization.” The trade data describes something entirely different.

By tracking every position open and close chronologically through the deal log, I determined the maximum number of simultaneously open positions at any point in the backtest:

27 concurrent open positions, peaking on March 14, 2024.

Here is what actually happened during that event. As gold declined from approximately $2,178 toward $2,153 over March 13–14, 2024, the EA opened a cascade of buy positions at progressively lower prices:

• Buy @ $2,178.19
• Buy @ $2,175.52
• Buy @ $2,173.44
• Buy @ $2,171.73
• Buy @ $2,170.01
• Buy @ $2,168.15
• Buy @ $2,166.33
• Buy @ $2,164.41
• Buy @ $2,162.42
• Buy @ $2,159.99
• Buy @ $2,158.21
• Buy @ $2,156.65
• Buy @ $2,153.56

…and more, until 27 positions were open simultaneously. Every position was tagged in the trade comments with grid-step identifiers like [T2/S03] and [T2/S04] — where “S” denotes the grid step level. The EA’s own parameter set even includes InpGridColor=65535, an explicit acknowledgment in the code that this is a grid system.

This is textbook grid averaging: as price moves against the position, open more positions at worse prices, lowering the average entry, and wait for a reversal to close the entire basket in profit. It worked in March 2024 because gold reversed. The fundamental question every buyer must ask is: what happens when gold does not reverse?

Why the 81% Win Rate Is Misleading

A grid system almost always produces a high win rate, and understanding why is essential to evaluating this EA honestly.

When a grid basket eventually closes in profit, every individual position in that basket is recorded as a separate “win” — even the ones that were deeply underwater for days. In the April 29, 2025 cluster, for example, the EA closed 20 positions simultaneously at the same price ($3,328.81). Ten of those closed as small losses and ten as small wins, netting just $15.16 total — but the trade log records 10 wins from that single grid unwinding event.

This is why the win rate of 81% is not evidence of predictive accuracy. It is an arithmetic artifact of grid mechanics. The system wins frequently in small amounts because it refuses to close losing positions until the grid recovers. The losses, when they come, come all at once — as the equity drawdown reveals.

The reported average win of $2.37 versus average loss of $3.06 confirms this. The wins are tiny. The expected payoff of just $1.16 per trade across 1,750 trades is razor-thin. This is not a high-conviction directional strategy — it is a high-frequency profit-harvesting grid that accumulates small gains while warehousing risk in open drawdown.

Quantum Athena EA Review: The 3.3-Year Backtest Problem

The backtest covers January 2023 to April 2026 — approximately 3.3 years. For a gold EA released in 2026, this is a conspicuously short test window, and the reason matters.

The 2023–2026 period was overwhelmingly a gold bull market. Gold rose from roughly $1,800 in early 2023 to above $3,500 by early 2026 — one of the most sustained directional advances in the metal’s history. For a grid system that opens buy positions on dips and waits for recovery, a persistent bull market is the single most favorable possible environment. Every dip eventually recovers because the underlying trend is relentlessly upward.

What the backtest conspicuously excludes:

• 2020’s COVID crash and the violent gold volatility that followed
• 2021–2022’s extended gold consolidation and drawdown, when gold fell from $2,075 to $1,615 — a 22% decline over many months
• Any sustained gold bear market or prolonged sideways regime

A grid system tested only through a bull market is being shown in its best possible light. The 2021–2022 period — when gold spent over a year grinding lower and sideways — is precisely the environment that breaks averaging systems, because the dips do not promptly recover and the grid keeps adding positions into a falling market.

A 3.3-year backtest on a strategy this sensitive to market regime is not sufficient evidence of robustness. It is a curated sample.

The Z-Score Warning: -3.46 (99.74%)

The MT5 report includes a Z-Score of -3.46 with 99.74% confidence. This statistic is frequently overlooked by retail traders, but it is highly informative.

A negative Z-Score indicates that winning and losing trades are not randomly distributed — specifically, that wins tend to follow wins and losses tend to follow losses, in streaks. A Z-Score of -3.46 at 99.74% confidence means there is a statistically significant tendency for the system to produce clusters of consecutive outcomes.

For a grid system, this is exactly what we would expect and exactly what creates danger. When the grid is working (market trending favorably), it produces long streaks of wins. When the grid is caught in an adverse move, it produces clusters of losses as multiple basket positions close together. The streakiness quantified by the Z-Score is the mathematical fingerprint of the concentrated risk events that the smooth balance curve hides.

The maximum consecutive loss event recorded was 10 trades for -$145.71 — and the single largest loss of -$104.14 on February 5, 2026 confirms that when the grid does unwind unfavorably, the damage is concentrated and significant.

Profit Concentration: The 2025 Anomaly

Breaking down net profit by year from the trade data reveals an uneven distribution:

The visual “PL by Year” chart shows 2025 generating nearly $994 — roughly half of the entire backtest profit in a single year. 2025 was the year gold went parabolic, surging from approximately $2,600 to well above $3,000. A grid system buying every dip in a parabolic uptrend is operating in its ideal environment, and the results reflect that.

The concern is straightforward: the strategy’s profitability is heavily dependent on the specific market conditions of 2025. In years with less directional momentum (which 2023 and 2024 partially represent, with roughly $475–$487 net each), the returns are far more modest. And in a genuine bear market — entirely absent from this backtest — a dip-buying grid faces existential risk.

Entry Timing: A Concerning Concentration

The “Entries by hours” chart reveals that the EA concentrates the vast majority of its position openings in a narrow window: hours 19, 22, and 23 (server time), with hour 22 showing over 540 entries. There is also a secondary cluster around hours 3–4.

This is significant for a gold grid system. Hours 22–23 server time typically correspond to the late US session rollover and the thin-liquidity window before the Asian session fully engages. During these hours:

• Spreads on XAUUSD widen substantially, even on ECN brokers
• Liquidity is reduced, increasing slippage on grid entries
• The daily swap/rollover is charged, and grid systems holding many open positions accumulate substantial financing costs

The backtest confirms the swap impact: across the full test, the system paid -$314.49 in swap charges and -$207.66 in commissions — a combined $522.15 in costs, equal to roughly 26% of the entire net profit. On a live account with wider real-world spreads during these thin hours, these costs would be materially higher, directly eroding the thin $1.16 per-trade expected payoff.

The “Quantum Queen Light” Lineage: What It Tells Us

Marketing the Quantum Athena as a “light version of the legendary Quantum Queen” is a deliberate positioning choice, and it carries information.

The Quantum Queen and related Quantum family EAs have generated mixed long-term results in the retail community. Grid-based gold systems within this family have, in various forms, experienced significant drawdowns and account losses when gold conditions turned unfavorable. The “light” designation suggests reduced grid aggression — fewer maximum positions, smaller step sizes, or tighter risk parameters — but the fundamental architecture remains grid-based, as the 27-position March 2024 event proves.

A “lighter” grid is still a grid. It defers and conceals risk in the same way; it simply may take a larger or more sustained adverse move to trigger catastrophic loss. The core vulnerability — that the system can accumulate a large basket of losing positions with no realized stop-loss until the market recovers — is unchanged.

The Long-Short Imbalance: 93% Long

The trade distribution shows an extreme directional bias: 93% long trades versus 7% short trades.

In a backtest period that was overwhelmingly a gold bull market, a 93% long bias is essentially a leveraged bet on gold continuing to rise. This is not a market-neutral or adaptive strategy — it is a system structurally positioned for one direction, validated against a period in which that direction was correct.

If gold enters a sustained downtrend, a 93%-long grid system faces the worst possible scenario: repeatedly buying into a falling market, accumulating ever-deeper baskets of losing long positions, while its minimal 7% short exposure provides almost no offsetting protection. The backtest cannot show this risk because the test period contained no such downtrend.

Risk Management Reality: What $699.99 Actually Buys

Let’s assess what a buyer is actually getting for the $699.99 price tag.

The position sizing: The backtest used InpLotsFixed=0.01 with auto-lot calculation tied to balance (InpAutoLotsValue=3, InpLotsFixedBalance=500.0). This means on a $1,000 account, the EA traded micro-lots. At larger account sizes or more aggressive auto-lot settings, the grid’s position count of 27 simultaneous trades would scale proportionally — and so would the equity drawdown. A 27.48% equity drawdown at 0.01 lot becomes the same percentage at larger sizes, but the dollar exposure multiplies.

The concealed leverage: With up to 27 open positions, the effective market exposure at the grid’s peak is 27x the nominal single-position size. The 1:500 leverage used in the backtest provides the margin headroom for this, but it also means a sufficiently large adverse gold move could trigger a margin call before the grid has a chance to recover.

The thin edge: An expected payoff of $1.16 per trade, with $522 of the $2,023 profit consumed by swap and commission, leaves a genuinely thin net edge. Any degradation in live execution — wider spreads, slippage on the thin-hour entries, higher real swap rates — could compress this edge significantly.

Quantum Athena EA Review: Professional Pre-Deployment Checklist

For any trader seriously considering this EA, here is the due-diligence checklist:

1. Understand that this is a grid system, not a precision strategy. Regardless of marketing language, the trade data confirms grid averaging with up to 27 concurrent positions. Price it as a grid system, with all the tail-risk that implies.

2. Focus on the equity drawdown, not the balance drawdown. The relevant risk number is 27.48%, not 7.25%. The balance curve’s smoothness is an artifact of deferred loss realization. Plan your capital around the equity figure.

3. Recognize the backtest excludes the conditions that break grids. The 2023–2026 period was a gold bull market. The 2021–2022 sideways/down period and the 2020 crash are absent. You have no data on how this system behaves in the conditions most dangerous to it.

4. Account for the swap and commission drag. $522 of costs against $2,023 profit is 26% of gross. On a live account with realistic spreads during the thin-liquidity entry hours this EA favors, expect higher costs.

5. Be cautious of the 93% long bias. This system is structurally a leveraged long-gold position. If your market thesis includes any probability of a sustained gold downtrend, this EA is positioned against you.

6. The product is brand new (April 2026). There is minimal live track record. The linked MQL5 signal should be examined critically for total duration, real drawdown, and whether it has operated through any adverse gold conditions — which, given gold’s 2025–2026 trajectory, it almost certainly has not.

7. Test at minimum position size for at least 6 months across varied conditions. A grid system must be observed through at least one significant adverse move before any capital scaling. Do not let a smooth opening few weeks create false confidence.

Final Verdict: Quantum Athena EA Review Without the Marketing Gloss

The Quantum Athena EA is a competently constructed grid system with genuinely strong backtest statistics under favorable conditions. Its 100% real-tick MT5 backtest methodology is rigorous, its win rate is high (as grid systems’ win rates always are), and its performance during the 2023–2026 gold bull market was legitimately profitable on paper.

But the honest assessment cannot end with the surface statistics.

This is a grid averaging system that opened 27 simultaneous positions during a single adverse move, that carries a 27.48% equity drawdown concealed behind a 7.25% balance drawdown, that was tested only across a 3.3-year gold bull market while excluding the sideways and bear conditions that destroy grid strategies, and that is structurally 93% long on an instrument it bets will keep rising. The Z-Score of -3.46 quantifies the streaky, concentrated nature of its risk. And $522 of its $2,023 profit was consumed by trading costs.

The Quantum family heritage — explicitly grid-based — is the most honest description of what this EA is. The “precision” and “intelligent optimization” framing in the marketing obscures the mechanical reality that the trade data lays bare.

For beginner traders: A high win rate means almost nothing when it comes from a grid system. The 81% figure is arithmetic, not accuracy. Understand that grids defer losses rather than avoid them.

For intermediate traders: The balance-vs-equity drawdown gap is the most important metric on the report. Whenever you see a 4x divergence between those two numbers, you are looking at a system that warehouses risk in open positions. That is the number that can blow your account.

For experienced algorithmic traders: A grid tested exclusively through a bull market, with a -3.46 Z-Score, 27 max concurrent positions, and a 93% long bias, is a regime-dependent bet on gold continuing to rise. Without a backtest spanning 2020–2022 and a verifiable live account through adverse conditions, the $2,023 profit is not evidence of robustness — it is evidence of favorable timing.

Grid systems can be profitable for extended periods. They are also responsible for a disproportionate share of catastrophic retail account losses, precisely because they look flawless right up until the move that breaks them. The Quantum Athena EA has not yet faced that move in any data available to a prospective buyer.

Trade with your eyes open.

Even more advisors with test results are presented in our advisor database.

In order to download an adviser with tests, go to our telegram channel 👇

Сообщение Quantum Athena EA Review: Shocking Hidden Grid Risk Exposed появились сначала на EA ForexLab.