CFA Level II ExamQuantitative MethodsMedium
A quantitative researcher is performing a backtest on a new trading strategy. The strategy involves using market data that was publicly available at the time of the trading decisions. However, the researcher inadvertently uses a dataset that includes data corrections and revisions that were only made public months after the original trading signals would have been generated. Which bias is most likely introduced into the backtest results?
- ASurvivorship bias
- BLook-ahead bias
- CData snooping bias
- DSelection bias
Show answer & explanationAnswer & explanation
Correct answer: B. Look-ahead bias
Look-ahead bias occurs when a backtest uses information that would not have been available to a trader at the time the trading decisions were made. Using revised data that was only available months later is a classic example of incorporating future information into past decisions.
Why the other options are wrong
- A. Survivorship bias occurs when only existing entities are included in the analysis, ignoring those that failed or ceased to exist.
- C. Data snooping bias arises from repeatedly testing strategies on the same dataset until one appears profitable by chance, not from using future information.
- D. Selection bias occurs when the sample data is not representative of the population intended to be analyzed, often due to how data is chosen.
Look-Ahead Bias
A bias in backtesting where a trading strategy is evaluated using information that would not have been available at the time the trading decision was made.
- Uses future information to make past decisions.
- Artificially inflates strategy performance.
- Common sources include revised financial data, future index constituents.
Memory trick: Don't peek into the future when testing the past!