Why we publish the backtests that failed (and why you should trust that)

A vendor who only shows winning EAs is selling a filter, not a strategy. We link every EA to its real 5-year backtest — including the ones that did not pass. Here is the honesty policy.

The filter problem

If an EA seller shows you five green curves, the only thing you have learned is that they know how to pick five green curves. The losing strategies are still on their drive — you just are not allowed to see them. That is not evidence of a good strategy. It is evidence of a good filter.

So we decided to do the opposite. Every EAbase EA links to its real backtest, and the ones that did not pass are shown as blocked or not-backtested, with the reason written next to them. No deleting. No renaming. No "that one doesn't count".

What "real" means here

A real backtest is five years of history (2020–2024) on a live broker dataset (Exness), with the actual trade log — not a three-month cherry-pick ending on a lucky streak. The numbers we show are the ones that survive scrutiny:

• Profit factor — gross profit divided by gross loss. Above 1.10 is workable; above 1.5 is solid.
• Net profit and max drawdown — together they tell you whether the edge is worth the worst-case pain.
• Trade count — below ~20 trades, the statistics are noise, not signal.

We do not show a single equity curve and call it proof. We show the test, the limits, and the failures.

The honest failures

Four of our multi-symbol EAs are currently marked "not backtested". The reason is not a bad strategy — it is that the trial account used for testing does not expose the secondary symbols those EAs need (a second currency pair, or the gold/silver leg). That is an account limitation, and we say so in plain text rather than hiding the gap.

About twenty other EAs produced lower or negative results in the same test. They are listed with their real numbers, not quietly retired. You can read the drawdown and decide.

Why this is good for you

A vendor who only shows winners is asking you to trust a filter. A vendor who shows the losers is asking you to judge the engineering. The second one is the only one worth your money, because the day your account hits a drawdown, you want to know the strategy was built to survive it — not dressed up after the fact.

The line we will not cross

No guaranteed returns. No fabricated live accounts. No anonymous "a client made 400%". When we publish a real buyer case study, it is verified and the buyer has signed off in writing. Until then, that section stays empty rather than filled with stories we cannot prove. See the standing policy on our proof page.