Recursive Factor Generation and Automated Factor Management
Summary
The document describes a framework for computing quantitative stock factors as recursive transformations of numerical indicators. It says factor logic can be defined with programming languages or XML: procedural languages suit more complex structures, while XML can express recursive rules for generating factors in batches. It also presents automated factor generation as a way to search for signals beyond manually designed factors.
The cited examples include 226 breakout factors, of which 20–30% reportedly showed promising cross-sectional stock selection; their performance fell after orthogonalization, suggesting overlap with traditional factors. A large-order buying turnover share factor derived from tick data reportedly performed strongly, and an equal-weight portfolio of orthogonalized factors achieved nearly 20% annualized long-short return with a return-to-drawdown ratio around 9. The document cautions that tick data is costly to process and the backtest covered only a short period starting in August 2015, so stability needed further monitoring. It also explains that automatically generating database tables and SQL operations supports centralized factor management.
Key ideas
- Factor calculations can be represented as recursive transformations from existing indicators to new indicators.
- Programming languages and XML can both describe factor computations, with XML suited to batch generation.
- Automated search produced breakout factors with reported cross-sectional promise, but orthogonalization reduced their performance.
- A tick-data factor based on large buy orders showed strong reported results, though the backtest period was short.
- Automated table creation and SQL operations can support centralized factor storage and management.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.