Random Forest Stock Ranking with Portfolio Optimization

Summary

This research summary describes ranking international liquid stocks with a random forest trained on technical features. Rather than forecasting exact returns, the model estimates the likelihood that a stock will outperform an average-return threshold. The ranking is used to form top and bottom portfolios, including equal-weighted long-short portfolios and optimized portfolios; the summary also mentions a lower-turnover alternative. The model is trained across varied market environments and is not repeatedly refit over time.

The reported study finds that ranked portfolios outperform bottom-ranked and randomly assembled portfolios, and that minimum-variance optimization improves the long-short Sharpe ratio. It reports robustness to portfolio size, regional restrictions, and individual outliers, and says a momentum-extended Fama-French model does not explain the excess performance. These are claims from the supplied paper summary, not independently verifiable results here. The summary notes that its high alpha excludes transaction costs, while turnover and shorting constraints may limit practical results; it does not include the full feature definitions or validation details.

Key ideas

• The model ranks stocks by estimated probability of exceeding an average-return threshold.
• Technical features feed a random forest trained on international liquid equities.
• Rankings are evaluated through top-versus-bottom portfolios and portfolio optimization.
• The summary reports improved risk-adjusted results for minimum-variance long-short portfolios.
• Reported findings exclude trading costs and may be constrained by turnover and short availability.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.