Predicting the Movement of the S&P 500 Index using Machine Learning

Sammanfattning: Predicting the stock market has been a longstanding topic of interest in financial research. It is regarded as a highly challenging but important task given the vital role the financial markets play in shaping the global economies. In this thesis, the goal is to predict the movement of the S&P 500 Index using machine learning methods. To this end, we apply two machine learning algorithms, random forest, and logistic regression, to financial data in a quest to try and predict if the S&P 500 Index will move in a positive or negative direction the following day. To further test the validity of the best performing machine learning model in our study, we develop a dynamic trading strategy where the predictions of the model act as an investment signal. If the model predicts that the S&P 500 Index will move in a positive direction the following day, we invest in equities (SPDR S&P 500 ETF Trust). Conversely, if the model predicts a negative movement, we instead invest in fixed income (Vanguard Total Bond Market ETF). We assess the performance of the trading strategy by comparing its Sharpe ratio to a second strategy, a traditional portfolio that holds 60% equities and 40% fixed income.