Analysis and Forecasting of Saudi Stock Market Using Time Series Algorithms

Forecasting the stock market is essential for investors to help assess the potential financial movement of the market value. Therefore, accurate estimation of the movement and price of the stock contributes more to the profit of investors through making informed investment judgments. Thus, the popularity of forecasting the stock market increased because of the difficulty of this field and that it is unforecastable and cannot be controlled because of its volatile nature and is affected by several factors. Nevertheless, the field of machine learning succeeded in forecasting the future stock market based on the analysis of time series data collected over time. The main objective of this paper is to develop day-to-day forecast models in ways that have been popular among researchers through time series data collected from the Saudi stock market, Tadawul, containing daily observations of the Tadawul All Share Index (TASI), and a different sample of some companies listed under the TASI. In our analysis, we relied on the technical approach, so it was important to choose technical indicators most closely related to the closing price of the next day using feature engineering and feature selection. Then, this study conducted two different models of Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) time series techniques. Best expected results Performance was evaluated using the Root Mean Square Error (RMSE) scale. The results show that LSTM algorithms based on deep learning are better than the ARIMA statistical model in stock market forecasting.