Submit your trading rules and get a free preliminary backtest. If you want the complete out-of-sample and robustness analysis, unlock the full verification.
A real run of our engine on this page load: buy when the 50-day SMA crosses above the 200-day, sell when it crosses back. Not a recommendation.
Every screen in the video is the product running real backtests. Discover and YouTube analysis are part of Pro.
Every backtest runs on the daily bars below, one bar per trading day, with the same execution rules. Coverage differs per stock, so each stock’s own dates are listed.
| Symbol | From | To | Bars | Last close |
|---|---|---|---|---|
| Loading… | ||||
A single historical backtest can look good by chance. The full verification tests whether it survives.
You pay for the verification, not the verdict.
A strategy can fail it, and the report will say so.
In TradingView: open a chart → Pine Editor → paste → Add to chart, on the daily timeframe. Its numbers can differ slightly from ours because its data and fill model are its own.
Our testing method grew out of an experiment: a simulated fruit-fly brain wired to stock prices. We locked the rules, tested it once on years it never saw, and published the results, including that it failed. The fly brain does not test or improve your strategy. Your backtests run on a plain, deterministic rule engine; the experiment is simply where this way of testing was proven in public.
The strategy is written down exactly: signals, sizing, costs.
The plan is written down before training: which data, which comparisons, which pass/fail thresholds.
Anything tunable is tuned here, and only here. The test years stay unopened.
Final settings are frozen and hashed, then the untouched years are run. Every run against a lock is recorded.
Random entries, benchmarks and the thresholds set in step 2 decide. Data already looked at is no longer unseen.
Loading the published results…
These are the fly-brain experiment's own recorded results, shown as published. They are not produced by the rule engine that runs your strategy.
A hash shows whether code, data or thresholds changed after the freeze. It cannot stop a test being run again: the run log is what makes repeat looks visible. Once a test window has been examined, it is no longer unseen data, and we say so rather than calling it out-of-sample twice.
| Analysis | Preliminary backtest | Full verification |
|---|---|---|
| Total return, win rate, trades, max drawdown | ✓ | ✓ |
| Buy & hold comparison over the same dates and costs | ✓ | ✓ |
| Equity curve, drawdown and price chart with your fills | ✓ | ✓ |
| Plain-English strategy input and TradingView script | ✓ | ✓ |
| Out-of-sample test on data the rules never saw | — | ✓ |
| Parameter freeze with hash, and a run log | — | ✓ |
| Robustness: nearby settings and random-entry controls | — | ✓ |
| Random-entry controls and rule-based benchmarks | — | ✓ |
| Fees/slippage sensitivity | — | ✓ |
| Detailed trade analysis | — | ✓ |
| Full verification report with verdict | — | ✓ |
Loading plans…