Hybrid fuzzy inference rules of descent method and wavelet function for volatility forecasting.

Area of Science:

• Financial econometrics
• Time series analysis
• Machine learning for finance

Background:

• Accurate stock market volatility prediction is crucial for financial decision-making.
• Traditional models often struggle with the complex dynamics of stock prices.
• Saudi Arabia's stock exchange (Tadawul) presents unique market characteristics.

Purpose of the Study:

• To develop and evaluate a novel hybrid model for improved stock market volatility prediction.
• To assess the impact of oil prices and repo rates on stock market behavior.
• To compare the proposed model's performance against established forecasting methods.

Main Methods:

• Utilized Maximum Overlapping Discrete Wavelet Transform (MODWT) with five mathematical functions and fuzzy inference rules (FIR.DM).
• Employed oil price (Loil) and repo rate (Repo) as input variables, with stock market price (LSCS) as the output.
• Applied multiple regression correlation and Engle-Granger Causality tests to validate input-output relationships.
• Implemented a hybrid MODWT-FIR.DM model, specifically MODWT-LA8-FIR.DM, for volatility forecasting.

Main Results:

• Input variables (oil price and repo rate) were found to significantly influence stock market prices.
• The hybrid MODWT-LA8-FIR.DM model demonstrated superior performance in volatility prediction compared to traditional FIR.DM and other MODWT-FIR.DM models.
• The proposed model achieved lower error metrics (ME, RMSE, MAE, MPE) on test datasets.

Conclusions:

• The hybrid MODWT-LA8-FIR.DM model shows significant potential for accurate stock market forecasting.
• The integration of wavelet transform and fuzzy logic enhances the prediction of stock market volatility.
• This approach offers a promising tool for investors and financial analysts in the Saudi stock market.