Building and Weighting a Multi-Factor Stock Selection Model
Summary
The document outlines a systematic stock-selection model built from valuation, profitability, growth, operating-efficiency, and technical factors. It describes a scoring approach: choose candidate factors, test their predictive usefulness, remove redundant factors, combine the remaining factors with weights, and refine the model. The premise is that factors with persistent historical ability to distinguish future returns may be combined to rank stocks.
A backtest covering 2011–2017 compares equal weighting with several approaches based on information-coefficient history. The report says equal weighting, a 12-month average IC method, and a 12-month IC-IR method performed well, with monthly win rates reported between 68% and 72%. Equal weighting is presented as the strongest overall approach; the 12-month average IC method is described as more vulnerable when market styles shift quickly, while IC-IR weighting is said to manage drawdowns better. These are historical results, not guarantees. The report flags macroeconomic deterioration and weakening or reversal of factor effectiveness as risks.
Key ideas
- A multi-factor model can combine valuation, profitability, growth, operating, and technical measures to rank stocks.
- A scoring model can be built by testing candidate factors, removing redundancy, and weighting the survivors.
- The report compares equal weighting and weighting schemes based on historical information coefficients.
- Historical backtests favor some weighting methods, but market-style changes can affect their results.
- Factor effectiveness may weaken or reverse, and historical performance does not guarantee future returns.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.