Automated trading with performance weighted random forests and seasonality
作者：
Highlights：
• We describe a model for seasonal stock trading using a novel ensemble of random forests approach.
• We explore the effectiveness of various regression techniques and methods for expert weighting.
• Random forests are found to produce the best out of sample results.
• Averaging the experts’ predictions weighted by their recent performance provides the greatest risk adjusted return.
摘要
•We describe a model for seasonal stock trading using a novel ensemble of random forests approach.•We explore the effectiveness of various regression techniques and methods for expert weighting.•Random forests are found to produce the best out of sample results.•Averaging the experts’ predictions weighted by their recent performance provides the greatest risk adjusted return.
论文关键词：Random forests,Ensemble learning,Stock price prediction,Machine learning,Algorithmic trading
论文评审过程：Available online 12 December 2013.
论文官网地址：https://doi.org/10.1016/j.eswa.2013.12.009
