One number, built from five attempts to break your strategy.
The Strata Score is a 0–100 composite of the robustness checks described in the methodology. It exists so you don’t fund a strategy on the strength of one flattering equity curve. Here is exactly what goes into it — weights and all.
The components
Net P&L relative to account size, Sharpe ratio, max drawdown, and profit factor on the historical run. If a prop firm is selected and the historical drawdown exceeds that firm's dollar limit, the drawdown portion is zeroed — a strategy that would have blown the account doesn't get partial credit.
Survival across 1,000 bootstrap resamples of the trade sequence. Measures whether the result depends on the lucky ordering of trades that happened to occur historically.
Performance on rolling walk-forward windows the strategy never trained on — by default six months in-sample, one month out, stepped across the full history.
Enough trades to mean something, spread across enough trading days, without one outsized day carrying the whole result. Thin samples score low no matter how pretty the curve.
Percentile rank of the strategy's P&L against up to 5,000 randomly composed strategies on the same data. Edge has to clear the random pile first.
Every score payload carries the raw value, weight, and weighted contribution per component, so the breakdown you see in the app is the actual math — not a simplified story about it. If a component can’t be computed, the score is flagged as degraded and names the missing pieces instead of quietly substituting an optimistic default.
Grades
| Score | Grade | What it means |
|---|---|---|
| 90+ | A+ | Survived everything we threw at it. Still not a guarantee. |
| 82–89 | A | Robust across reshuffles and unseen data. |
| 74–81 | B+ | Solid, with a weak spot worth reading the components for. |
| 65–73 | B | Promising. Check out-of-sample stability before sizing. |
| 50–64 | C | Marginal. Likely some curve fit in the result. |
| 35–49 | D | Fragile. The robustness checks found real problems. |
| Below 35 | F | Discard. The edge didn't survive testing. |
What the score is not
It is not a profit prediction. It measures how a strategy held up under historical stress — reshuffled, held out, and compared against random — and nothing about markets promises the future will resemble the past.
It is not investment advice. It’s a research instrument: the same strategy can score differently against different prop-firm rule sets, because surviving a $2,000 trailing drawdown is a different problem than surviving a $4,500 one.
And it is not for sale. Strategies are scored by the same code path regardless of who built them or what plan they’re on.
Futures and derivatives carry substantial risk. Backtest results do not guarantee future returns. Strata is software — not investment advice.