Adaptive Machine Learning for Stock Market Monitoring

With the thriving of research on machine learning and the demand for innovative methods of approaching stock markets, recent years have seen a surge of applying machine learning techniques to stock markets. This dissertation explored the effectiveness of some basic machine learning models applied to the stocks from the Information Technology Sector of the S&P 500 Index over a period of five years (from October 1st, 2013 to September 30th, 2018). And these application were under a time adaptive model training framework that was proposed to simulate the trading scenarios in real world. After comparing the prediction performances of different models based on the ``prediction matrices'', it seemed that random forest (RF) models were more accurate than simple MLP and LSTM models when predicting price trends of the target stock by the other stocks in the IT Sector, although RF models may suffer from low betting frequency during the trading simulation.Mathematics, Department o