Skip to content

Catch overfitting before it costs you

A great backtest on one slice of history proves almost nothing — markets change, and a strategy tuned to the past often falls apart on the future. The Robustness tab exists to tell you, before you risk a cent, whether your edge is real or just curve-fit.

Monte Carlo

Shuffle and resample your trades thousands of times to see the range of outcomes your edge could produce — not just the single lucky (or unlucky) path you happened to backtest. You get a cone of equity curves and the odds of hitting various drawdowns. If the median path is healthy and the bad tail is survivable, that's a good sign.

Walk-forward (WFO)

The honest way to test a strategy with parameters. StrategyNodes optimises on a window of history, then tests on the next, unseen window — and rolls that forward across all your data. A strategy that only looks good when it can see the whole dataset will fall apart here. One that survives walk-forward has earned some trust.

Cross-market retest

Thanks to <Primary>, you can re-run the exact same strategy on other symbols and asset classes in one click. An edge that only exists on EURUSD H1 is suspicious; one that holds up across several markets is far more believable.

Stability & MAE/MFE

  • Parameter stability — does the strategy still work if you nudge the inputs? A peak that collapses the moment you change the period from 20 to 21 is overfit.
  • MAE / MFE — how far trades go against you (Maximum Adverse Excursion) and in your favour (Maximum Favourable Excursion) before closing. Great for tuning stops and targets.

The StrategyNodes Score

The Score rolls all of this into one honest number (0–100) and a grade (A–D), with sub-scores for robustness, generalisation and risk. It runs your strategy through a holdout it has never seen, against a secret benchmark, and applies honesty checks that penalise curve-fit, look-ahead and over-trading.

Think of it as a second opinion you can't game: a high Score means the edge survived scrutiny it didn't get to study for. You can even share a Score as a public page (your strategy's logic stays private — only the result is shown).

How to read it

A strong backtest with a weak Score = probably overfit. A modest backtest with a strong Score = a smaller but more real edge. Chase the second one.

Confident it holds up? Take it live → Exporting to Pine & MQL5.