Write the experiment first
Record the hypothesis, asset universe, dates, data fields and signal timing. Decide which result would challenge the idea. Keep a benchmark that uses compatible dates and return conventions.
Model more than entry signals
Specify orders, fills, position sizes, fees and slippage. Inspect a small set of trades by hand before reading aggregate results. Record how corporate actions and missing observations are handled.
Read the report in context
Review the equity path, drawdown, exposure, turnover and distribution of outcomes. Preserve the configuration and a later evaluation period. A historical result is not a guarantee that the same rule will perform similarly in future markets.
Sources & references
References support specific definitions or methodological context. Examples and research checklists are explanatory; no performance result is claimed.