Time Series Model for Stock Market Prediction Utilising Prophet

Abstract

The importance of predicting the rise and fall of the stock market is indisputable, as it helps investors make wise decisions in terms of buying and selling stocks; however, underlying nuances of the stock market make it difficult to build accurate prediction models. Time is an important facet in determining stock trends, which unfortunately is always neglected. Our approach highlights the significance of time to improve the accuracy of the prediction, which is done by utilising the prophet library. In order to improve the accuracy, the dataset[2] was thoroughly investigated and visualised by constructing numerous graphs which shed light on how time has impacted the stock prices. The prophet library defines three hyperparameters namely seasonality, trend, and holidays. Each of these parameters elucidate the gravity held by time in stock market prediction. We have employed the dataset [2] of a multinational financial services company called the Banco Santander, S.A., this dataset[2] contains the stock prices of the Santander Group, which is Located in Spain. Moreover, it is the 16th largest banking institution in the world. It was hypothesised that data which is nearer to the data to be predicted held much more significance as compared to historic data and this hypothesis was validated, indicating the importance of time based data for an accurate prediction.  The month of May saw a drastic increase in the stock price of our dataset[2] and a slight increase on Thursdays.