Modeling Stock Volatility in Saudi Stock Exchange Using Classification Approach
Abstract
Stock volatility is an aspect that has been studied by numerous scholars, businesses, and
investors with the primary aim of prediction. With accurate predictions, the investors are likely
to plan on when to purchase/buy or sell the stock shares. The objective herein is to fit some
regression models on companiesǯ stock and evaluate the model fit. The study selected nine
companies. Three of the companies were small, three medium, and the other three were large
size company. Given that Saudi Arabia is an oil-based economy, it was more likely to show the
effects of oil price fluctuations. The companǯs selection was important in determining the
effect of oil price fluctuations, Fedǯs rate, credit rate, and covid-19 pandemic on different levels
of the economy. The first phase included fitting linear regression, where it was established that
all the suspected factors had a significant effect on stock value. The oil prices and credit rate
were found to positively affect the stock prices in large, small, and medium companies. On the
other hand, covid-19 and Fed funds rate generally had a negative effect on the stock values.
The logistic, Naïve Bayes, and random forest were adopted to predict the stock volatility on
whether the stock would go up or down. Although the predictor variables were insignificant.
The model accounted for at least 45% of the total variation in stock variation. The largest
impact of the predictor variables was seen among the small companies. The naïve Bayes had
the highest accuracy suggesting that this was the best fitting classification algorithm