Quantitative Research Questions on Factors, Overfitting, and Strategy Design
Summary
This page collects questions for a quantitative trading discussion rather than supplying answers. Topics include the Sharpe ratio, overfitting and the gap between backtests and live results; outlier treatment and avoiding look-ahead bias when building factors; and methods for choosing weights in multi-factor models. It also asks how to measure crowding in price and volume factors and find less correlated sources of alpha.
A further question concerns feature selection for rolling XGBoost models. The poster observes that a factor with lower absolute information coefficient can sometimes improve returns and reduce drawdown more than a factor with a higher information coefficient, and asks how to build a feature library efficiently. Other questions raise the possible use of large language models to screen convertible bond announcements when data updates are delayed, and request feedback on a proposed research workflow. Since the page contains prompts without responses, it offers no methods, evidence, or conclusions for resolving these issues.
Key ideas
- The page raises questions about Sharpe ratios, overfitting, and the gap between backtests and live trading.
- It asks how to handle outliers, prevent look-ahead bias, and set multi-factor weights.
- It discusses factor crowding and the search for less correlated alpha sources.
- It questions whether information coefficient alone can guide feature selection for rolling XGBoost models.
- The page poses these topics but provides no answers or supporting evidence.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.