Every file was run in MetaTrader 4 on EURUSD H1, 32 months of history, 17 points (1.7 pip), measured on the live feed. Nothing here is inferred from a vendor's screenshot. The conditions are published next to the numbers so you can re-run them yourself.
Of the 77 files: 34 never opened a position, 20 opened positions and closed none, 8 were stopped out without a clean trade record. Only 21 produced a closed-trade record at all. All of them are published.
Take the five EAs with the best raw profit and delete their best 5% of trades. 4 of the five go negative. That is not bad luck — it is the shape of a recovery grid.
| EA | ■ full sample ■ after removing the best 5% of trades | P/L | After tail cut |
|---|---|---|---|
| Gold Hedge | +$6,888 | −$1,878 | |
| Hedge Grid | +$5,758 | −$2,158 | |
| Bi-Dir Grid | +$2,541 | −$1,400 | |
| Hurricane ECN 2.0+ | +$1,418 | +$803 | |
| Smart Hub | +$1,385 | −$40 |
Why the cut is set at 5%. A system whose profit depends on a handful of its best trades is not sensitive to its own entry logic. It is sensitive to whether those few trades happened to appear in the sample. Re-run the same EA over a different window and the profit can simply not be there.
Four failure modes account for almost the whole batch. None of them are visible on a vendor's equity screenshot.
Every one of these started at 0.01 lots and finished holding between 1.4 and 5.8 lots. The loss was not caused by a bad signal; it was caused by the size the system assigned to a losing sequence.
One file opened 87,580 positions and closed zero. Without a close there is no trade record — only an equity line, which is exactly what is sold to you.
Positive in the full sample, negative once the best 5% is removed. These are the ones that look best in a marketing post and are the least likely to repeat.
Indicators wrapped as EAs, copy-trade plumbing, and files that need manual inputs nobody supplies. They pass a file scan and fail a real test.
Positive after the tail cut, with at least 100 closed trades and at least $100 left standing. This is the entire list.
| EA | Verdict | Opened | Closed | Win rate | P/L | P/L w/o best 5% | Lots |
|---|---|---|---|---|---|---|---|
| Hurricane Scalper ECN 2.0 PLUS ea066 | Survived | 5,099 | 583 | 86.1% | +$1,418 | +$803 | 0.01→0.83 |
| Hurricane Scalper ECN ea069 | Survived | 5,099 | 583 | 86.1% | +$1,418 | +$803 | 0.01→0.83 |
| Hurricane Scalper Optimised ea065 | Survived | 2,536 | 392 | 76.5% | +$1,111 | +$729 | 0.01→0.03 |
| Hurricane Scalper ECN-A (Cent) ea068 | Survived | 935 | 298 | 100.0% | +$268 | +$216 | 0.01→0.05 |
Two of these four files produce byte-identical results — the same EA under a different name. So that is one strategy, tested three ways: the plain ECN build, the optimised build, and a cent-account variant. Not three discoveries. Three packages.
Seven EAs in total clear the tail cut, but three of those have fewer than 100 closed trades and two more finish under $10 — a sample that cannot support a verdict either way. They are in the database as Survived, with their trade counts showing.
Every EA gets a page whether it worked or not. We publish the conditions, the trade count, the lot escalation and the result after the tail cut. We do not publish "expected profit", we do not rank by return, and we do not accept payment to change a verdict. If a result is sponsored or a vendor paid for the test, the page says so at the top.