Historical Testing Overview
Category indexHistorical dataNot predictive
Overview
Historical testing shows how user-defined rules behaved on available past data using stated assumptions.
Not predictive
Historical tests are diagnostic. Costs, slippage, liquidity, missing data, latency, and broker behavior can affect future outcomes.
Guides in this section
Test setup
Configure dataset scope, dates, capital, position rules, and test assumptions.
Assumptions
Document data, costs, slippage, and execution assumptions.
Metrics
Read returns, drawdown, turnover, win/loss, and exposure metrics carefully.
Limitations
Understand why historical reports cannot validate future behavior.
Testing workflow
Strategy schemaData ingestionSimulation engineReport reviewNext experiment
Assumptions table
| Assumption | User check |
|---|---|
| Costs | Brokerage, fees, taxes, and slippage. |
| Data quality | Missing candles and instrument mapping. |
| Execution | Candle model versus live order behavior. |
| Risk | Position sizing and loss controls. |
Common pitfalls
- Overfitting to one period.
- Ignoring missing candles.
- Understating costs and slippage.
- Treating AI summaries as decisions.
- Exporting before reviewing risk settings.