Signals of a useful backtest
Good validation gives you reasons to trust or reject the process. Bad validation gives you a screenshot.
- Do not tune until the bad period disappears.
- Do not compare systems with incompatible capital assumptions.
- Do not treat one high-performing run as robust evidence.
Make the environment believable
Date range, instrument set, and capital base should resemble the desk you plan to run rather than the dataset that flatters the strategy most.
The equity curve should be readable
A strong terminal profit can still hide intolerable drawdowns, unstable streaks, or fragile dependence on one period of market behavior.
A pass means worthy of monitoring, not guaranteed
Backtesting narrows uncertainty. It does not remove slippage, changing regimes, liquidity shifts, or execution drift in live conditions.