Strategy creation and backtesting
Tuned onTested on months it never saw
- Strategy builder
- Your own EA
- Backtester
- Optimisation
- WFO
- RWFO
Cookies
We use cookies the site needs to work. With your OK, we’ll also use them to count visits and to see which of our ads bring people here. You can change this any time from the footer. Cookie settings
Build a strategy, backtest it on tick data and walk it forward on months it never saw. Then combine the ones that hold up into one portfolio and follow its live trades. No code to write.Build a strategy, backtest it on tick data and walk it forward. Combine the ones that hold up into one portfolio and follow it live. No code to write.Build a strategy, backtest it and walk it forward. No code to write.
Neighbour checkFast MA × Slow MA
Each square is one setting. The most profitable one is a lone spike, far brighter than its neighbours. The pick sits on a broad plateau, where nearby settings do about as well.
Rank 1 by profit
+47,313Walk-forward17/145
10 of 12 periods held up
Neighbour score0.86
Fast MA 17, Slow MA 145. Ranked 15th by profit, and its neighbours do about as well.
| Rank | Fast/ | Profit | Neighbours | Walk-forward | Verdict |
|---|---|---|---|---|---|
| 1 | Fast 41, Slow 260 | +47,313 | 0.31 | 3 of 12 periods held up | Spike |
| 2 | Fast 9, Slow 85 | +39,108 | 0.24 | 2 of 12 periods held up | Spike |
| 3 | Fast 33, Slow 215 | +34,216 | 0.38 | 4 of 12 periods held up | Spike |
| 4 | Fast 52, Slow 120 | +31,902 | 0.44 | 5 of 12 periods held up | Spike |
| Ranks 5 to 14 not shown13 ranks between | |||||
| 15 | Fast 17, Slow 145 | +24,851 | 0.86 | 10 of 12 periods held up | Broad plateau |
0.86
Neighbour score
0.75 and up: broad plateau
OptimisationRolling walk-forwardMA cross · 2,500 runsMA cross · 12 windows
Held up
| Rank | Fast / Slow | Profit | Walk-forward | Verdict |
|---|---|---|---|---|
| 1 | 41 / 260 | +47,313 | 3/12 | Spike |
| 2 | 9 / 85 | +39,108 | 2/12 | Spike |
| 3 | 33 / 215 | +34,216 | 4/12 | Spike |
| 4 | 52 / 120 | +31,902 | 5/12 | Spike |
| 15 | 17 / 145 | +24,851 | 10/12 | Plateau |
| Window | Tested on | Picked | Out-of-sample | Result |
|---|---|---|---|---|
| 9 | Mar to Apr 2026 | 17 / 150 | −880 | Lost |
| 10 | May to Jun 2026 | 18 / 145 | +2,730 | Held up |
| 11 | Jul to Aug 2026 | 17 / 145 | +3,280 | Held up |
| 12 | Sep to Oct 2026 | 17 / 140 | +2,690 | Held up |
| Next | From Nov 2026 | 17 / 145 | Trade next | Trade next |
An optimisation tests thousands of settings. Sort by profit and the top row looks like an edge. But test enough settings and one will look great by chance alone. With five years of data, about 45 truly different tries are almost sure to turn up a good-looking winner that has no edge at all.An optimisation tests thousands of settings. Test enough of them and one will look great by chance alone: with five years of data, about 45 truly different tries are almost sure to turn up a winner with no edge at all.Test enough settings and one will look great by chance alone.
Bailey, Borwein, López de Prado and Zhu, “Pseudo-Mathematics and Financial Charlatanism”, Notices of the AMS, 2014.
| Rank | Fast MA | Slow MA | Profit |
|---|---|---|---|
| 1 | 41 | 260 | +47,313 |
| 2 | 9 | 85 | +39,108 |
| 3 | 33 | 215 | +34,216 |
| 4 | 52 | 120 | +31,902 |
| 5 | 16 | 140 | +29,908 |
| and 2,495 more rows | |||
Example data, made up for illustration.
Bailey, Borwein, López de Prado and Zhu, “Pseudo-Mathematics and Financial Charlatanism”, Notices of the AMS, 2014.
Four modules, in the order you use them.
Tuned onTested on months it never saw
Drag to turn it. Example data, made up for illustration.
Also on any single run: Monte Carlo, Prop Firm, What-If, Swap Stress, Trade Explorer.
Strategywright holds no money and places no trades. The live link is a small reporting add-on that sends your results to the app, and nothing in it can open an order.Strategywright holds no money and places no trades. The live link only sends your results.
Nothing in it can open an order
Free trial
Analyst
For traders who backtest in MetaTrader 5 and bring the results here.
Join the waitlistFull suite
Prices in US dollars.
No. Build a strategy from ready-made rules, or describe it in plain words. If you already have an EA, upload its MQL5 file.
MQL5 source files (.mq5), to backtest an EA you already have. MetaTrader 5 optimisation exports, saved as CSV, TSV, XML or ZIP, for the Report Analyzer.
No. It holds no money and places no trades.
Yes. Every tool reads it except five checks that compare each setting with its neighbours. Those need most of the grid tested, and say so when a genetic run skipped too much.
No tool can promise that. It tells you whether a strategy holds up on months it never saw, or was most likely luck.