What a Trading Bot Backtest Proves (and What It Hides)

A backtest runs a strategy's rules against historical prices to simulate what it would have earned. It's the standard way to sanity-check a strategy before risking real money.

Backtests catch real problems: a strategy that loses on every historical period is probably bad, and parameter sensitivity (performance collapsing when you tweak one number) reveals overfitting.

What backtests hide: survivorship isn't an issue for price data, but execution is — a backtest assumes fills at historical prices that may not have been available (slippage in thin markets). Fees are often undercounted. And past conditions never repeat exactly — a strategy tuned to a 2021 bull run may bleed out in 2026 chop.

The overfitting trap: optimize parameters until the backtest looks great and you've likely fit noise, not signal. A strategy that works 'okay' across many time periods beats one that's perfect on exactly one.

Use backtests to filter out bad strategies and understand behavior — not to predict returns.