Abstract

The information technology (IT) field has continued to witness a series of innovations and advancements. The stock market sub-sector has also witnessed a series of large capitalisation, it is now a hub of increasing investment for investors. Based on this, many analysts, investors, and domain experts have developed a sequence of statistical tools to predict the outcome of the market. However, the stock market has witnessed the transition from traditional statistical tools of stock prediction to an era of algorithmic trading built on the foundation of artificial intelligence.

The artificial intelligence approach has greatly improved the accuracy of stock market prediction but remains one of the most challenging tasks because the market is categorised to be volatile. The technical indicators for prediction like Open Price, Low Price, High Price, and Closed Price is not static with Time. Hence the outcome of the data is usually non-linear in a relationship. The purpose of stock market prediction is to predict the future values of a company's stock. Recent prediction technologies have been developed to perform this task using machine learning by making a prediction using the past and current values of stock to predict the future values through a training method. Machine learning uses different models to make prediction easy and simple.

This thesis is focused on deploying a classification model Random Forest Classifier, and Sentiment Analysis as the base model for stock market trends while a Regression Model Long Short-Term Memory for prediction of next day close price for ten global brands (Companies). The Random Forest Classifier was designed to provide insight on the direction of the market to advise buyers/traders if to buy or not and also advise the users how confident is the system for every prediction knowing well that the market is very volatile. The Long Short Term Memory was deployed to learn the dynamics of long-term price for making price predictions into the future because the model has a capacity to store long-term data which is very important knowing that stock market prices can be voluminous and released per second. A third level decision support system was deployed to guild users using sentiment analysis. The sentiment was considered a factor that provides insight knowing the power of speculations and news headlines within the stock market. The strengths and weaknesses of applying artificial intelligence tools were also discussed so as the associated threats and opportunities with this particular study. The unique addition to the body of knowledge using artificial intelligence algorithms in stock market prediction was considered.

Artificial Intelligence Algorithms for Stock Market Prediction