Prediction of stocks performance in Bursa Malaysia using fuzzy logistic regression / Azri Farhan Mahammad Hanif, Batrisyia Kiranah Shaiful and Nadhilah Saleh

While financial ratios are currently the most well-known method to understand and predict the performance of a firm and their stock, there is a lack of explaining which of those ratios are significant to the prediction. In previous studies, the significance of financial ratios varies with industries and countries, but there were none able to illustrate which of these variables significantly influence the performance of a company. In fact, that logistic regression is suitable for predicting binary outcomes, while stock performance is better expressed linguistically using "very good, good, average, bad, very bad". Thus, this study will help to highlight significant financial ratios that could be used as a guideline for investors to make an informed decision. Multicollinearity test is used to remove financial ratios that are highly correlated to one another, and fuzzy logistic regression is used to predict the stock performance of top 30 companies in Bursa Malaysia. The ratios which were deemed significant was chosen to develop the fuzzy logistic regression model and a comparison between the model and classical logistic regression model was made. It was found that the fuzzy logistic regression model did not out-performed classical logistic regression model in aspects that was measured. This may be due to the small number of observations and the nature of the financial market which is noisy and non-deterministic.