Fitness Objectives for Genetic Programming of Equity Factors
Summary
This reference describes several candidate fitness objectives for genetic programming that produces equity factors. It defines IC information ratio as the mean information coefficient divided by its standard deviation, with the information coefficient measuring the relationship between factor values and subsequent stock returns. It also names mutual information as an objective, though it gives little detail about its calculation.
The remaining measures evaluate daily factor-sorted portfolios. Stocks are divided into ten groups; the long-short series uses the return difference between the highest and lowest groups, while the long-only series uses the highest-factor group. Each series can be scored by cumulative summed returns, Sharpe ratio, or annualized volatility. These choices emphasize different properties, such as predictive association, risk-adjusted performance, or return variability. The document is a glossary rather than an empirical comparison: it supplies no data, results, implementation details, or guidance on costs, turnover, constraints, or out-of-sample validation. Those omissions matter when selecting a fitness function for strategy search.
Key ideas
- IC information ratio scores the mean factor-return correlation relative to its variability.
- Mutual information is listed as a fitness objective, but its calculation is not explained here.
- Daily factor sorts create long-short returns from the top and bottom groups and long-only returns from the top group.
- Return sums, Sharpe ratios, and annualized volatility offer different objectives for evaluating the sorted portfolios.
- The reference does not compare objectives or address trading costs and out-of-sample validation.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.