Intraday Market Return Predictability Based on the Factor ZOO
Log in to collectAcademic paper
Intraday Market Return Predictability Culled from the Factor Zoo
Saketh Aleti; Tim Bollerslev; Mathias Siggaard
- Duke University
- ?Duke University, Department of Economics
- National Bureau of Economic Research
- ?Duke University - Department of Economics
- ?Duke University - Finance
- ?National Bureau of Economic Research (NBER)
- DKAarhus University
Strategy in a nutshell
The strategy uses high-frequency intraday returns for 272 portfolios and over 200 risk factors. LASSO regression identifies relevant predictors, and trading models forecast 15-minute market returns. Trades are executed based on predicted returns, with strategies including long-short, long-only, and signal-strength-based approaches. Performance is evaluated via intraday returns and Sharpe ratios, consistently outperforming a buy-and-hold benchmark.
Economic rationale
The rationale is to exploit short-lived predictability in high-frequency market returns. By leveraging machine learning on extensive factor data, the strategy captures information not immediately reflected in prices, improving risk-adjusted performance and market timing efficiency.