DDrawLedger
← All guides

Analysis

Backtesting: testing strategies honestly

A backtest replays a strategy against history to see how it would have done. Done honestly — no future data, a random baseline, out-of-sample checks — it is the only way to evaluate a strategy at all.

The rules of honest backtesting

Three rules matter most. No lookahead: decisions use only data available at the time. A baseline: a strategy is only interesting if it beats random selection, because random selection alone matches a few numbers per draw by luck. And out-of-sample discipline: if you tuned parameters on the same history you test on, the test is invalid.

What the backtesting lab shows

Every strategy on this site — frequency, cold, gap, recency, pairs, balanced — is run under those rules against a seeded random baseline. On a fair lottery, the honest result is that none of them beat chance.

Keep reading