How to verify trading bot performance claims

How to verify trading bot performance claims before paying: backtests vs live results, verified Myfxbook feeds, drawdown checks, and seller red flags.

30-day free trialCard requiredCancel before renewal

Start free trial
All insights
EducationAutomationRisk

How to verify trading bot performance claims

How to verify trading bot performance claims before paying: backtests vs live results, verified Myfxbook feeds, drawdown checks, and seller red flags.

You can verify a trading bot’s performance claims in three moves: insist on live results rather than backtests or demo, insist those results are independently verified — a Myfxbook or FX Blue feed with its verification checks passed, or read-only investor access to the account — and insist on seeing the full history: every losing trade, the worst drawdown, and the risk settings that produced the numbers. Anything less — screenshots, hand-picked months, win rates quoted without context — is marketing material, not evidence.

That is the whole method. The rest of this guide unpacks why each part matters: how backtests get curve-fitted to the past, why demo results flatter live execution, how sellers hide drawdown and dead accounts, and what genuinely honest reporting looks like. None of it needs trading experience — only the willingness to ask for things a vendor with real results can supply in minutes, and a vendor without them cannot supply at all.

Not all evidence is equal

Performance evidence comes in grades, and most of what circulates in EA marketing sits at the bottom of the scale. Before you examine any specific claim, place the evidence you have been shown on this ladder — it tells you how much weight the claim can possibly carry.

EvidenceWhat it provesWhat it can hide
Screenshot of a platformAlmost nothing — trivially faked, cropped or stagedLosses, dates, whether the account even exists
Backtest reportHow the logic behaved on past data under the tester’s assumptionsOverfitting, unreal spreads, zero slippage, a hand-picked window
Demo accountThe software runs and places tradesReal spreads, slippage and rejections — demo fills are kinder than live
Live account, self-reportedReal money was traded at some pointSelective disclosure — you see the account the seller chose to show
Live account, third-party verifiedA broker-confirmed trade history over a stated periodSurvivorship across multiple accounts — and the future
A rough hierarchy of trading bot performance evidence, weakest first.

The practical rule: base your decision almost entirely on the bottom row, and treat everything above it as background detail at best.

Backtests: useful to the builder, weak for the buyer

A backtest replays a strategy over historical prices. Builders need them — they are how ideas get tested at all — but as buyer-facing evidence they carry a structural flaw called overfitting. A strategy with enough adjustable parameters can be tuned until it fits past data almost perfectly, the way a curve can be bent through any set of points. The tuning captures the noise of that particular history rather than a repeatable edge, and the performance evaporates on data the strategy has never seen.

  • Curve fitting: dozens of optimisation runs against one dataset will always produce a flattering equity curve. Ask whether the parameters were fixed before the test period, and whether out-of-sample or walk-forward results exist.
  • Execution assumptions: a fixed narrow spread and zero slippage can flip a losing strategy into a “profitable” backtest — on a fast symbol like gold the gap is substantial.
  • Data quality: MetaTrader backtests range from real tick data down to interpolated one-minute bars, and the lower grades invent prices that never actually traded.
  • Window selection: a report covering only one strong trending year says nothing about the chop before or after it. Ask why the window starts and ends where it does.

Demo results flatter live execution

A demo account trades imaginary money against the broker’s quoted prices, so every order fills instantly at the price on screen. Live accounts trade against real liquidity: spreads widen around news and rollover, stop orders slip in fast markets, and fills can be requoted or partial. On a volatile symbol such as XAUUSD those effects are large enough to erase a thin edge — which is why a strategy can look consistently profitable on demo and lose money live on identical settings.

Demo evidence is still worth something: it proves the software runs, respects its own rules and manages orders correctly. It is just not profitability evidence, and a seller presenting demo results without labelling them as demo has already answered your honesty question.

What “Myfxbook verified” means — and what it doesn’t

Third-party tracking services such as Myfxbook and FX Blue connect to a trading account and republish its history. On Myfxbook, two separate checks matter: “track record verified”, which confirms the published history is complete and untampered (usually via read-only investor access), and “trading privileges verified”, which confirms the trades were actually placed by that account rather than imported. Look for both, check the account is marked live rather than demo, and confirm the history contains no hidden or private periods — sellers can hide individual trades or date ranges, and a gap is a statement in itself.

Now the limits. Verification authenticates the past of one account; it says nothing about the future, and nothing about the accounts you were not shown. A vendor can run ten accounts on ten different settings and publish the one that survived — survivorship bias in its purest form. So ask the follow-up questions: how many accounts run this strategy, do the published settings match the product actually being sold, and has this exact configuration traded through both quiet and violent markets?

Red flags in an EA track record

  • A win rate quoted alone, with no average win-to-loss size — the single most common trick in EA marketing (more on this below).
  • No losing trades, days or months anywhere in the material.
  • Percentage gains with no risk context — “40% in a month” is meaningless without knowing the drawdown risked to get it.
  • A track record that starts suspiciously recently, or has gaps, hidden trades or private periods.
  • Screenshots and video “proof” where a verified feed or investor access would cost the seller nothing to provide.
  • Guarantees of any kind — guaranteed profit, a guaranteed prop-firm pass, “risk-free” anything. Real trading offers none of these, and a seller who promises them is telling you how they operate.

The martingale smoothness trap

Be especially suspicious of equity curves that climb in a near-perfect straight line with a very high win rate. That signature often belongs to martingale or grid systems, which increase position size after losses. The account banks many small wins and rarely closes a loser — the losses accumulate silently as floating drawdown on open positions instead — until one sustained move against the grid wipes out months of gains, and sometimes the account. On a track record, check the open-trade (floating) drawdown figure, not just the closed equity curve: a smooth line sitting above deep floating drawdown is the classic martingale fingerprint.

Win rate means nothing without risk-reward

A win rate only becomes information when it is paired with the average size of wins and losses. The arithmetic is unforgiving:

Win rateAverage win / lossResult over 100 trades
90%Wins $10, loses $15090 × $10 − 10 × $150 = −$600
40%Wins $200, loses $10040 × $200 − 60 × $100 = +$2,000
Illustrative arithmetic only — a high win rate can lose money and a low one can make it. Always ask for both numbers.

The checklist: what to demand before paying

Ask for a verified live recordA Myfxbook or FX Blue link with the relevant verification checks passed, or read-only investor access you can open in your own platform. Refusal is itself an answer — an honest seller loses nothing by providing this.
Confirm the account type and conditionsLive, not demo. Note the broker, the symbol traded and the leverage, and ask whether those conditions resemble the account you would actually run the bot on.
Read the full date rangeLook for an unbroken history spanning different market conditions — trending and ranging, calm and violent. Be wary of records only a few weeks long, or that begin right after an obvious strategy change.
Find the lossesLocate the losing trades, the losing weeks and the maximum drawdown — including floating drawdown on open positions. If you cannot find any of them, stop here.
Demand the risk contextRisk per trade, the lot-sizing method, and whether the published account ran the same settings being sold to you. A record produced on settings you cannot buy proves nothing about the product.
Interrogate the win rateAsk for the average win and the average loss alongside it, then run the hundred-trade arithmetic above yourself.
Trial it before money changes handsRun the bot on a demo or small account and compare its live behaviour against the published record. A seller who discourages trialling is telling you what the trial would show.

What honest reporting looks like

  • Losses shown by default — losing trades, losing periods and the worst drawdown appear in the headline material, not buried three clicks deep.
  • Methodology stated: symbol, timeframe, risk per trade, and the spread and slippage assumptions behind any backtest.
  • Continuous, dated history with no hidden periods, on an account type that is clearly labelled.
  • Risk framed honestly: past results are never presented as a promise, and no outcome is guaranteed.
  • A way to verify before paying — a verified feed, investor access, a genuine trial, or all three.

This is the standard we try to hold ourselves to. RSForex Bot publishes its trading results with the losing trades and losing periods included, and offers a 30-day free trial specifically so you can watch it trade before paying anything. That is not a claim of superiority — trading involves real risk, losing stretches happen to every strategy, and no software can guarantee an outcome. It is simply the level of evidence you should demand from us, and from anyone else selling automation.

Risk disclosure. Trading involves risk. RSForex Bot does not guarantee profits, account growth, or prop-firm outcomes. Users remain responsible for their own broker, prop-firm, account settings, and trading decisions. Past performance does not guarantee future results.

Myfxbook and FX Blue are independent third-party services, named here for education only and unaffiliated with this site. Their verification confirms historical account data as reported by a broker; it is not an endorsement of any product and says nothing about future performance.

Share X LinkedIn
Free trial

Put disciplined gold automation on your MT5.

RSForex Bot runs the XAUUSD strategy locally inside MetaTrader 5 on supported FTMO and OANDA accounts, with built-in per-trade, daily-loss and drawdown limits.