We proved the bot before we built the sales page.
Most trading bots are sold on screenshots and promises. Ours comes with three years of results on two markets, every losing month included. The rules were tested on data they had never seen, then on a second market they were never built for. They held on both.
We share the evidence here. We don't share the code, the rules or the settings; that's the product. The Opening Range Bot releases in the coming weeks.
Why we publish this
- A highlight reel of winning days.
- A backtest fitted to the same data it's judged on.
- No drawdowns, no losing months, no sample size.
- A new system every few months, none of them proven.
- Three years of results, every trade, every losing streak.
- Two years of data kept closed until the very end, then used as the verdict.
- The same rules run unchanged on a second market.
- Twenty-six variations tested. Most failed, and we say so.
Everything we tested
Twenty-six variations across ES and NQ, grouped by what we changed. Five held up on data they weren't tuned on. Everything else was dropped, including ideas that looked best at first. We publish the count, not the details.
| What we varied | Tested | Held up | testedheld up |
|---|---|---|---|
| Entry method | 4 | 2 | |
| Range length | 2 | 0 | |
| Filters | 3 | 0 | |
| Bar size | 3 | 2 | |
| Risk limits | 3 | 1 | |
| Exit management | 3 | 0 | |
| Session timing | 5 | 0 | |
| Entry cut-off | 3 | 0 | |
| Total | 26 | 5 |
"Held up" means positive on both the research data and the data held back, by a margin we'd trust. Variations that only matched the released version are counted as failed alternatives.
Account growth, trade by trade
Version 1 is the bot as first built. Version 2 is the release, after one refinement found in testing. We kept Version 1 on the chart on purpose: it shows how close a good-looking idea can come to losing money, and why we test before we sell.
The rough patches, shown in full
Every year, and every side
An edge that only works in one year or one direction is usually luck. Version 2 made money in every full year, on long trades and on short trades, on both markets.
2023 is the last two weeks of December only.
What a single trade looks like
ES and NQ side by side
The rules were settled on ES first. NQ was tested afterwards with the same rules, so every NQ session was new to them. NQ moves more dollars per contract, so every number scales up, losses included. Micro contracts make either market accessible on a smaller account.
MES and MNQ are one tenth the size, with about $1 commission per round trip. Their figures are scaled from the full-size results.
The same bot, running in ATAS
This is the bot on a live NQ chart in ATAS, in simulation mode. It replays three years of history and shows the result on its dashboard: 585 trades, 52% winners, profit factor 1.31, +$158,945. That matches our research to the dollar, because it's the same code.
The standard a bot has to pass here
This is less about one bot than about the bar we hold every system to before it reaches you.
- Real exchange dataContinuous ES and NQ futures, three years, checked for real volume every month.
- The bot's own code, not a spreadsheetThe exact engine that trades in ATAS replays the history. Inside ATAS the same code matches the research to the dollar.
- Data held backTwo of the three years stayed closed during research. They decided the verdict. Two ideas that looked best early on failed there and were dropped.
- A market it wasn't built forThe rules came from ES. They ran unchanged on NQ and held.
- Stress testsSlippage, commissions, timing and risk limits. A result that only works at one exact setting isn't trusted.
- Replay, simulation, then liveUnder way now: Market Replay with real order handling, then weeks on a simulated account, before release.
What we're not sharing
You've seen every result. You haven't seen how the bot decides when to trade, where it places its stop and target, or how it manages a winner. That stays private. Publishing results without the recipe is how we can be this open and still have something worth buying.
What you'll get
- Runs inside ATAS
- Add it to an ES, MES, NQ or MNQ chart. It reads the instrument and applies the settings that were tested for it.
- Protection at the broker
- Every trade goes in with a stop and a target resting at your broker, so it stays protected even if your computer goes down.
- Risk limits you control
- A daily loss cap and an optional per-trade limit, so it fits your account or your prop firm's rules.
- Prove it yourself first
- A built-in simulation mode replays your own chart history and shows the results on a live dashboard before a single order is sent.
Reading the numbers
- Profit factor
- Money won divided by money lost. 1.00 breaks even. 1.30 means $1.30 made for every $1 lost. Day-trading systems that hold up usually sit between 1.2 and 1.5.
- Win rate
- Share of trades that made money. About half win here; it works because the average win is larger than the average loss.
- Drawdown
- Not a single trade. It's the worst stretch from an account high to the low that followed, over the whole three years. Size the account for it.
- Luck test (t)
- How likely results this good are by chance. Under 1 is easily luck; about 2 is unlikely to be luck. ES scores 2.6 and NQ 2.8.
Opening Range Bot for ATAS
ES, MES, NQ and MNQ. Built on three years of tested results, released only after replay and simulation checks. Get notified at zerodaytrades.io.