On my public Myfxbook profile there's a system called NeuralPath. Its showcase numbers: +641.24% gain, 32.86% drawdown, 81% win rate, over nine thousand trades. If NeuralPath were for sale with a landing page and three testimonials, it would fly off the shelves.
Now I'll teach you to dismantle it in five minutes. The exercise is worth double: first, because 90% of the systems sold on the internet have exactly this physiognomy; second, because it's mine — and if you learn not to trust my showcase numbers, you're vaccinated forever against everyone else's. NeuralPath is an experiment I keep public on purpose: a didactic crash-test dummy.
The analyst's five minutes
Minute 1 — The account type
First line of the page: Demo. Not real. A demo track record proves nothing about live execution: no real slippage, instant fills even where the real market would have no liquidity, laboratory spreads. The analysis could stop right here. But let's continue, because the best part comes later.
Minute 2 — Average win and average loss
Average win: 31.66 pips. Average loss: −168.43 pips. W/L ratio ≈ 0.19: you risk more than five units to bring one home. With R = 0.19 the break-even win rate is 1/(1+0.19) = 84%. The system wins 81%. It sits below structural break-even: the whole mountain of small wins exists to pay for the rare enormous losses, and it almost never suffices. This profile — win often and little, lose rarely and enormously — is the statistical signature of hidden tail-risk strategies: grids, martingales, averaging down. The win-rate mountain is the bait; the tail is the hook. (Check any system with the break-even win rate calculator.)
Minute 3 — Pip expectancy
Myfxbook prints it: −5.3 pips per trade. Negative. How does a system with negative pip expectancy show +641% in dollars? Because size isn't constant: the system wins pips when it's light and loses pips when it's heavy. The % gain is inflated by compounding and size progression, not by an edge per unit of risk. When the two metrics diverge like this — splendid dollars, negative pips — the explanation is always in size management. And never in a good way.
Minute 4 — The Z-score
−6.12. The Z-score measures dependence between consecutive outcomes: near zero = independent trades; strongly negative = wins arriving in correlated series — exactly what a grid closing in clusters produces. Six standard deviations aren't a hint: they're a written confession. The trades aren't independent bets; they're installments of the same mortgage.
Minute 5 — Average duration and density
Average duration 2 minutes, nine thousand trades in a few months, and best/worst trades (+$8,491 / −$3,322 on a $10,000 account) that alone move 30–80% of the starting capital. Everything converges on the same diagnosis.
The permanent checklist
Generalized, for any track record — mine included. Five pairs of numbers, none of which is the gain:
- Real or demo? And if real: verified with trading privileges? For how long? How many accounts not shown (survivorship)?
- Avg W / avg L, and the break-even win rate that follows, compared with the actual win rate. 90% of scam systems die at this line.
- Expectancy in pips/R as well as in currency. Divergence between the two = size manipulation.
- Z-score and average duration — the signature of grids and martingales.
- Max drawdown and its date: if the DD is "small" but the system has never crossed a hostile regime, it's small for lack of testing, not merit. And the final question: who pays this edge, and why should they keep paying? If the answer doesn't exist, the track record is just a lucky trajectory awaiting refutation.
None of these checks takes more than a minute on a public Myfxbook page. The fact that almost no EA buyer performs them is the reason the miracle-EA market exists.
The interactive tools to run them right now, on any system: the Debunking page. And to understand why even honest backtests lie: Overfitting: how to lie to yourself with style.