A solution to portfolio optimisation based on random forest and long short-term memory networks

The randomness of the stock market presents a significant challenge in utilising known information to construct investment portfolios that maximise returns while minimising risks. This paper employs the entropy weight method combined with the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to assign weights to indicators and subsequently ranks and selects stocks from the CSI 300 Index constituents in the Chinese market and the NASDAQ-100 Index constituents in the US. market. Using random forest (RF) and long short-term memory (LSTM) to predict stock closing prices, the experimental results show that the LSTM model achieves higher predictive accuracy and more stable errors. By selecting the top 7 stocks based on monthly returns to construct a portfolio, the study analyses the returns of investment strategies under different prediction models. The results demonstrate that the portfolio constructed using the LSTM prediction model outperforms other portfolios in terms of cumulative return, annualised return, and Sharpe ratio.