Markowitz Mean-Variance Portfolio Optimization with Predictive Stock Selection Using Machine Learning

With the advances in time-series prediction, several recent developments machine learning have shown that integrating prediction methods into portfolio selection is a great opportunity. In this paper, we propose novel approach to formation strategy based on hybrid model combines convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM) with robust input features obtained from Huber’s location for stock Markowitz mean-variance (MV) optimal construction. Specifically, study first applies method preselection ensure high-quality inputs formation. Then, predicted results are integrated MV model. To comprehensively demonstrate superiority of proposed model, used two models, equal-weight (1/N) LSTM, BiLSTM, CNN-BiLSTM, employed them as benchmarks. Between January 2015 December 2020, historical data Stock Exchange Thailand 50 Index (SET50) were collected study. The experiment shows stocks can improve performance, show they outperform comparison models terms Sharpe ratio, mean return, risk.