Чек-лист проверки бэктеста

Question and rules

• State one falsifiable research question before running the test.
• Record the exact strategy version, asset, timeframe, signal timing, exits, sizing and benchmark.
• Record parameter exploration and why the final rule was selected.

Data

• Record source, fetch date, timezone, resolution, start/end, gaps, duplicates and excluded bars.
• Verify OHLC consistency and distinguish a missing funding value from zero.
• Confirm entry uses information available at that historical time.

Execution and costs

• Include entry and exit fees using dated, venue-specific assumptions.
• Include spread, slippage, funding, partial fills and market impact where applicable.
• Document order types, bar sequencing, stop/target conflict handling and liquidation limits.

Validation

• Keep a chronological evaluation period untouched while choosing rules.
• Inspect multiple regimes, drawdown, trade count, loss concentration and sensitivity.
• Try higher costs and perturbed parameters.
• Record inconclusive and failed tests as well as favorable ones.

Interpretation

• Separate historical simulation from paper observation and real fills.
• State that backtests are not forecasts or evidence of a future edge.