Intraday backtests are sensitive to timestamp errors, missing bars, and unrealistic execution assumptions. Use this checklist before trusting a strategy report.
Timestamp and Calendar Checks
Confirm timezone, daylight-saving treatment, bar boundaries, holidays, and shortened sessions. Mixing UTC and Eastern Time can shift signals and orders.
Missing Data and Corporate Actions
Investigate duplicate timestamps, gaps, trading halts, stock splits, dividends, and symbol changes. Document whether prices are adjusted or unadjusted.
Realistic Execution
A bar close is not automatically an executable price. Include commissions, spread, slippage, latency, and market impact where relevant.
Prevent Look-Ahead Bias
Signals, normalization, model training, and parameter selection must use only information available at the decision time. Test multiple symbols and market regimes and reserve a final out-of-sample period.
See our price data guide and 1-minute datasets.
Historical backtests do not guarantee future results.