Spotting fake EA results: the patterns that give manufactured curves away
ConfirmedTrades team · 13 April 2026 · 8 min read
Backtest curves, demo accounts, cherry-picked windows, hidden baskets and martingale staircases: how each looks, why it works, and which checks expose it.
Fake does not always mean forged
Outright forged statements exist, but most misleading Expert Advisor results are technically true. They are backtests presented as live results, demo accounts presented as real, one good window presented as the whole history, or a strategy whose losses simply have not arrived yet. Each of these has a recognisable shape, and once you know the shapes they are hard to un-see.
1. The backtest in disguise
A curve that is too straight. Backtests run on historical data with perfect fills, no slippage, and often a fixed spread far tighter than anything real. The result is a near-linear upward line with tiny, regular dips. Real live curves are lumpy, because real fills are lumpy.
The tell on ConfirmedTrades is simple: a backtest cannot be live-synced. If a record is not at least live-synced, it was uploaded, and you have no way of knowing which terminal produced it. Insist on the trust label before you look at the shape.
2. Demo presented as live
Demo servers fill at the requested price, rarely slip, and sometimes quote spreads no real client gets. Scalpers and news traders look heroic on demo and ordinary on real money. Every account page on ConfirmedTrades shows the account type, and demos are flagged clearly. A forex robot seller who only has demo results to show is telling you something.
3. The cherry-picked window
A screenshot of three good months, with the six bad ones before it cropped out. Or an account that was reset, re-funded and restarted after a blow-up, so the history begins on a clean slate.
Two things defend against this. The drawdown ledger keeps the worst balance and equity drawdown per broker login, so deleting and reconnecting an account does not hide a past fall; the worst figure reappears in the Glance hub. And the monthly calendar shows every month since the account was connected, so a cluster of reds surrounded by greens is visible rather than cropped. See maximum drawdown explained for how the ledger works.
4. The martingale staircase
A martingale doubles or increases lot size after each loss, so that the eventual winner recovers everything. The balance curve looks like a smooth staircase with a very high win rate and hardly any closed losses. Then one day the run of losses is longer than the account can fund, and the whole thing goes.
The AI review runs two specific checks over the closed trades. First, whether lot size tends to rise immediately after a loss in chronological order. Second, the signature of a high win rate paired with an average loss far larger than the average win. Either one raises a high-severity flag. A strategy type of Martingale in the review means the pattern is strong enough that the review is confident.
5. The hidden basket (grid)
A grid opens several positions on the same symbol at intervals and closes them together when the basket turns positive. Like martingale, it shows a pleasant balance curve while carrying a large floating loss. The difference from martingale is that lots may not increase, so it evades the lot-progression check.
It does not evade the equity stream. Because the connector sends equity, not just closed trades, the growth index dips when the basket is underwater, and Max DD (equity) in Glance records how deep it went. The AI review also looks for baskets of three or more same-symbol trades closed within the same minute and raises a grid flag when that is a large share of trading.
6. No stop-losses
An automated trading system with no stop-loss never closes a loser on purpose. Its closed-trade statistics look clean right up until the day a position is closed by the broker on a margin call. The review reports the share of trades that carried a stop-loss and a take-profit, and raises a high-severity flag when fewer than about one in seven trades had a stop on a record of any meaningful length.
Note that some honest strategies manage risk with mental stops or basket equity stops rather than per-trade stops. The flag is a reason to ask, not a verdict. But the answer needs to be convincing.
7. Too few trades, too little time
Thirty trades in a month from any trading bot can produce almost any curve by chance. The review attaches a short-record flag to small samples, and the Score's longevity component rewards age and trade count. How long before a track record means anything covers the statistics. The short version: a pretty curve with a small sample is not evidence of anything yet.
A fast screening routine
- Trust label at least live-synced, ideally broker-verified with trading privileges.
- Account type Real, not Demo or Contest.
- No high-severity flags in the AI review; read the stop-loss percentage.
- Compare Max DD (equity) with Max DD (balance). A big gap means held losers.
- Calendar: how many months, how many red?
- Trade count and age: enough to mean something?
Do it in one place
On ConfirmedTrades, every published account shows its verification badges, an AI strategy analysis that flags martingale, grid and missing stop-losses, the full drawdown and risk stats, and execution costs per symbol — so you can vet a strategy before you trust it, or prove your own.