To Develop a Financial Data Analytics and Prediction System by Combining Multiple Data Sources

The Prediction of Stock market has been and will always be the goal of the investors, since Stock market plays a vital role in country's economy. Various theories and prediction methods have been proposed and debated from the very start as to how the prediction has to be done. Technical indicators predict the future price levels or general direction of the price levels by evaluating the past patterns. There has been lot of research going on about stock market prediction in areas like computer science, business, engineering, statistics etc. The goal of this thesis is to develop financial data analytics by generating Moving Average (MA) model with a better Sharpe ratio and Average Directional Index(ADX) and incorporate with the Textual Analysis of the stocks. In this thesis, the generated MA model was tested on 6 stocks. The first step is to find the strength of the trends for Close prices and apply MA and then accuracy of the model. The second step is to perform the Sentiment Analysis on the textual data containing News and 8-k's, which helps to find out whether the sentiment of text is positive or negative. It helps to find how the stock trend varied with the text data. Finally, both analysis are combined and accuracy is measured. When compared the results, by applying Moving Average and smoothing the prices along with consideration of strength of the trends more accuracy is gained. Also, when textual news is added to the technical analysis the strength of the trends is added to the textual features and found the accuracy how the stock trends are affected in accordance with the news.

The accuracy of the system increased when the Moving Average is applied along with the news articles. The average accuracy for technical analysis is around 62% and textual analysis is around 66% showing that the news articles have impact on the stock price trends.