Caution notes for traders.

Here's the thing worth understanding: a backtest isn't a red flag by itself. Every EA has to start somewhere, and no one can wait five years to publish any evidence at all — that would be an absurd bar, and it would leave you with nothing to evaluate for years. Backtests are a normal, necessary part of the evidence. Some of the most rigorous evidence we can offer is backtest evidence, tested across multiple brokers and multiple years, and we say so directly.

What actually matters is whether the backtest itself is methodologically honest. Was it run on real tick data, or generated/interpolated data that fills gaps with guesses? Does it disclose its “history quality” percentage, and is that percentage high? Was it tested on more than one broker's data feed, or does it live or die on a single, unverifiable source? A backtest that's been stress-tested across independent brokers and years is a genuinely different thing from one run once, on one feed, with parameters that happen to fit that exact feed perfectly.

A very high win rate — 85%, 90%, higher — sounds like exactly what you want. But a win rate on its own tells you nothing about risk. What tells you something is putting it next to the size of a typical win versus the size of a typical loss.

If the win rate is extremely high and the average loss is meaningfully larger than the average win, that's a specific, recognizable shape. It usually means a system is finding ways to close small, marginal trades in the “win” column — by holding a losing position longer, or by adding to it — rather than taking a clean, predefined loss and moving on. That approach can produce an impressive-looking win-rate streak for a long time. It's also, structurally, the same shape that produces the rare but severe drawdown when the market finally moves too far or too fast for the recovery mechanism to work. The high win rate isn't the risk. The size of the rare loss is.

This is genuinely checkable from any report a seller publishes. Look at “average profit trade” next to “average loss trade.” If the ratio looks upside down relative to the win rate, ask why, specifically — not rhetorically.

These numbers exist to describe how much return a strategy produced relative to how much risk it took on the way there. They're useful precisely because genuine, sustainable systematic strategies tend to cluster in a fairly consistent range — and numbers that sit far outside that range, sustained for years, are unusual enough that experienced allocators treat them as something to investigate, not something to celebrate. A Sharpe ratio in the double digits, held over a multi-year window, is the kind of figure that should prompt a closer look at how it was calculated and what data it's drawn from — not automatic excitement.

When a retail EA advertises numbers like that, there are really only a few honest explanations: the backtest window was chosen or shaped to fit unusually favorable conditions, the strategy is quietly taking on a form of risk that hasn't shown up yet in the sample, or the statistic is being calculated in a way that doesn't match how it's normally defined. None of those are disqualifying on their own — but “the number is dramatically higher than the rest of the field” is a real, specific thing to notice, not just an intuition to brush aside.

Watch for phrasing like “does not use martingale in its recommended default configuration” or “does not rely on unlimited recovery cycles.” These are carefully constructed sentences. They can be entirely accurate and still leave real room for the underlying mechanism to exist in the code, available to be switched on, present in a non-default mode, or active in a way the plain-English summary doesn't quite rule out.

The more useful question isn't “does the marketing say no martingale” — it's “has anyone independently verified that, against the actual behavior, not just the product description.” If a seller can't point you to that kind of verification, that's worth knowing before you decide how much to trust the claim.

A results table is easy to generate. A results table that's been tested by someone with no financial interest in the outcome — across more than one broker, ideally with the methodology published so it can be checked — is a genuinely different category of evidence. If a product's entire case rests on charts and numbers that only the seller has ever produced, that's not automatically dishonest, but it does mean you're the one taking their word for it.

Scarcity counters, countdown-style pricing, and results presented in bold celebratory colors are marketing techniques, not evidence. None of them are inherently dishonest — a genuinely limited launch allocation is a real thing, and we use one ourselves. But it's worth noticing when a page is working harder to make you feel like you're about to miss out than it is working to show you something you could verify. The two aren't mutually exclusive, but when a listing leans heavily on the first and lightly on the second, that's information too.