Optimal stock investment strategy using prediction models

Stock price prediction has traditionally been known as a challenging task. However, recent advancements in machine learning and deep learning models have spurred extensive research in predicting stock returns. This study applies these predictive models to U.S. stock data to forecast stock returns and develop investment strategies based on these forecasts. Additionally, the performance of the model-based investment strategy was compared with that of a widely recognized method, market capitalization-weighted investing. The results indicate that, overall, market capitalization-weighted investing outperformed model-based investing. However, the highest returns were observed in the model-based strategy. It was also found that model-based investing exhibits higher volatility in returns, with significant disparities between years of high and low returns. While investing through machine learning methodologies may be attractive to investors seeking high risk and high return, market capitalization-weighted investing is likely more suitable for those desiring stable returns.