Stock Price Predictions for FAANG Companies Using Machine Learning Models

Abstract: The financial industry is one of the highest grossing sectors in the world as it is estimated to represent 24\% of the global economy. As most companies want their asset value to increase, it is of high interest to make good investments which will increase in either the short or long run. The main aim of this thesis was to reveal the performance on predictions using two different machine-learning models, namely Random Forest and Artificial Neural Networks. The target variable that our models aimed to predict was the closing prices of stocks for the FAANG companies, all of which are traded on NASDAQ. Our models used data sets dated from 2010 until 2020, that included several different features that often are subject to technical, fundamental and macroeconomic analysis. As we used the year of 2020 as validation data, stocks were highly affected by the Covid-19 pandemic, that caused severe fluctuations in several sectors and of course the financial markets. This might have been the main reason why Artificial Neural Networks was more effective in predicting the closing price, since it took noisy processes into consideration. We believe though that the global pandemic made an impact on our predictions, that did not perform efficiently enough to use in investment decisions. However, a series of results concerning statistical properties of the models are of interest.