Modeling financial ratios of Malaysian plantation stocks using Bayesian Networks

This project explored the potential of Bayesian Networks in modeling financial ratios of Malaysian plantation stocks. Financial ratio is a crucial part of fundamental analysis which involves analyzing a company's financial statements to gain insights on its financial condition, future performance and aid decision-making in stock investment. Meanwhile, Bayesian Network is a popular graphical and probabilistic tool that is useful for reasoning under uncertainty in real life situations. The prospect of using Bayesian Network to model financial ratios in fundamental analysis is therefore explored in this project in the context of the Malaysian plantation industry. Using the local stock market data, relationships among financial ratios and stock price changes are modeled in a Bayesian Network, which will then be used to provide assistance in stock buying decision by predicting the stock price changes given observations of financial ratios. The network's reliability is tested on real data based on the counts of correct predictions and also comparison of portfolio mean return to that of a random portfolio. The resulting network is found to be reasonably reliable as it consistently predicted price changes correctly more than half of the times and also outperformed the random portfolios.