Backtestslie.
We ran 800+ of the market's most popular EAs on real ticks. Only a handful survived — and most of those were phantoms, merely fitted to a past they already knew. This is the record of that test.
The Funnel
How “thousands-of-percent-per-year” vanishes
We downloaded the popular market EAs from the MQL5 ranking and backtested each one year-by-year on real ticks (Model4), not demo data. Every year resets to $10,000 (simple, non-compounding) to strip out the illusion of compounding. The result looked nothing like the sales pages.
Out of 806 EAs real-tick tested, 0 made it to the end. Read the real build date out of each product binary and all five safety-bar survivors were compiled after the test window — built by someone who already knew the answers. The higher up the ranking, the more they cluster into gold EAs, and most of those break on real ticks. “Best-selling” and “genuine” are not the same thing.
Why they lie
The two mechanisms behind the lie
Behind every dazzling result sit the same two traps. Both make the past look beautiful while failing in the future.
Coarse demo data makes losers look like winners
Most sales results are built on coarse bar data and ideal spreads. Run the same EA on real ticks (true price action, real spread) and about 70% of the top sellers hit 90–100% max drawdown — the account melts. That smooth equity curve was an artifact of coarse data.
EAs built AFTER the test window
Check the build date of the product binary and most were compiled after our test window (2020–2023). Fit a curve to a past whose answers you already know, and it will always look beautiful. That is not proof of edge. All five of the safety-bar survivors were this phantom. Worse, the product page won't tell you: the “updated” date MQL5 returns is a copy of the original publish date, not the current version's. A product listed in 2017 can still ship a binary compiled in 2026. We read the compile timestamp embedded in the binary itself — across our 1,146 downloads, 70% were 2026 builds.
So we stopped using a fixed test window. We score every EA only on the period after the binary was built — the stretch its author could not possibly have seen. That is the only way to guarantee a backtest is genuinely out-of-sample.
Profit × Safety
Plotting them by return and safety
Vertical axis: annual return. Horizontal: max drawdown (how much unrealized loss you must endure). Up-and-left means “big return, small risk” — the ideal spot. Gold dots are our EAs (all measured on real ticks, every number public); gray dots are the market EAs that cleared our real-tick safety bar (PF≥1.3, effective DD≤40%) — though by build date, all six are in-sample. The shaded band on the right is the blow-up zone where most market EAs sink on real ticks.
Annual Return × Max Drawdown
Ours = measured real-tick (public, 14-yr basis) / Market = cleared the real-tick safety bar (all in-sample by build date)
Market EAs shown anonymously as a methodology record (product names withheld). Beyond the blow-up band, at DD 90–100%, sits the majority of the real-tick-tested market EAs (the melted-account group).
Our lineup
Our lineup (measured real-tick, fully public)
Every published EA of ours shows only measured MT5 real-tick numbers — no flashy demo figures. Below, sorted by risk-adjusted return (annual ÷ max DD), pulled from the site's single source of truth.
| EA | Annual | Max DD | PF | Risk-adj. |
|---|---|---|---|---|
| ATLAStop | +68.8% | 19.7% | 1.48 | 3.49 |
| BITCOIN COMETtop | +57.6% | 18.2% | 2.19 | 3.16 |
| BITCOIN COIL | +45.0% | 21.4% | 3.59 | 2.10 |
| TRINITY | +97.8% | 69.0% | 1.79 | 1.42 |
| MEGAMAX DON EURJPY | +22.0% | 17.7% | 1.51 | 1.24 |
| ETHEREUM EA | +16.9% | 14.0% | 3.98 | 1.21 |
| BITCOIN GLACIER | +45.8% | 47.0% | 2.40 | 0.97 |
| MEGAMAX DONCHIAN | +39.3% | 58.6% | 1.55 | 0.67 |
Risk-adjusted = annual ÷ max DD (higher = safer growth). All MT5 real-tick (mt5_official). Latest values on each EA's detail page are authoritative.
Why we don't lie with numbers
Real-tick testing the world's popular EAs taught us one inconvenient fact: the numbers on sales pages can barely be trusted. So we measure every EA on real ticks and publish the result — good numbers and bad. We use build dates to keep verification honest, and we drop EAs that break. We choose surviving 14 years over looking flashy.
Backtests do not guarantee future results. Our figures here are measured MT5 Strategy Tester values (mt5_official), in sync with each detail page. Market-EA figures are results of our own real-tick tests, not the products' advertised numbers. Individual product names are withheld as a methodology record. Trade at your own risk.