The Walk-Forward Checklist

12 checks that separate a real edge from a curve-fit fantasy. The minimum bar every strategy should clear before you risk a dollar. Free. 4-page PDF.Most backtests lie. A beautiful equity curve is easy to fake and worthless live. This is the honest floor — run any strategy against all twelve before you trust the Sharpe on your screen.WHAT'S INSIDEBefore you backtest- Hypothesis first: write the economic reason the edge exists before touching data- Split before you look: hold out the final slice until the end- Freeze the rules: every parameter fixed up frontLook-ahead & leakage- No future data in features- Trade the next bar's open, not the signal bar's close- Point-in-time data onlyThe walk-forward itself- Roll, don't single-split: optimise on window N, test on N+1- Ten or more out-of-sample folds; report the distribution- A real edge shows up across most folds, not one lucky windowCosts & reality- Realistic transaction costs: commission, slippage, spread- Capacity and liquidity: could you fill that size?- Multiple-testing penalty: deflate for the number of variants triedThe one-line test: "Would this survive if I'd never seen the test set?" If no, you have a story, not a strategy.When you're ready for the code, the Honest Backtesting Kit turns this checklist into working Python. See recommended products below.

Found on Devadex — the discovery index for independent software the big search engines bury. More from gumroad.

Report this listing