What are common biases in backtesting?
Quant ResearchLevel: Intermediate
What are common biases in backtesting?
Hints
- When are data available, signals formed and trades executed?
- Check costs, latency, turnover and how many hypotheses preceded strategy selection.
Answer
Important biases include look-ahead, survivorship, repeated strategy selection, favorable-period selection and unrealistic fills or costs. Use point-in-time data with publication and revision timestamps, historically valid universes and delisted securities. Model executable prices, latency, fees, impact and position constraints. Separate hypothesis development, validation and a final holdout; keep a record of all trials. Paper trading can reveal implementation gaps, but it does not remove selection bias or prove future profitability.
Source and licence: Quant Research Interview Notes — SoYuCry; English translation by photon2563 · MIT · answer: Quantoros editorial