Stock picking with machine learning

Abstract We analyze machine learning algorithms for stock selection. Our study builds on weekly data the historical constituents of S&P500 over period from January 1999 to March 2021 and typical equity factors, additional firm fundamentals, technical indicators. A variety models are trained binary classification task predict whether a specific outperforms or underperforms cross‐sectional median return subsequent week. trading strategies that invest in stocks with highest predicted outperformance probability. empirical results show substantial significant learning‐based selection compared an equally weighted benchmark. Interestingly, we find more simplistic regularized logistic regression perform similarly well complex models. The robust when applied STOXX Europe 600 as alternative asset universe.