Random Forest Stock Ranking with Return Targets and Rank-Weighted Allocation

Summary

This brief strategy note outlines an equity selection approach using a random forest regression model. It adds features such as beta and technical indicators, including MACD and Williams %R, and predicts a return-style target calculated from the difference between selling and buying prices relative to the buy price. The model is described as using 15 trees with a maximum depth of 25 to limit overfitting.

The backtest approach gives larger allocations to stocks ranked more highly by the model and includes a stop-loss when a holding falls 3%. It offers two example allocation curves that concentrate capital increasingly in the top-ranked names, but reports no performance figures or detailed test design. The note is an outline rather than a reproducible research report: it does not define the prediction horizon, feature timing, training and validation scheme, trading costs, or portfolio rebalance rules. Those omissions make it impossible to assess data leakage, risk-adjusted returns, or whether the stated model constraints actually reduce overfitting.

Key ideas

• The approach ranks equities with a random forest regression model rather than a classifier.
• Its features include beta and technical indicators such as MACD and Williams %R.
• The target is the price return between buying and selling, expressed relative to the purchase price.
• Portfolio capital is weighted toward higher-ranked stocks, with a 3% stop-loss described.
• The note provides no backtest results or enough methodological detail to reproduce or evaluate the strategy.

Tags

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