The Research Department of Economics and Management of Tadbir Nikan Business, Marketing, and Finance Open 3092-6238 In Press 2027 08 01 Forecasting Weekly Returns of the Tehran Stock Exchange Total Index Using the MIDAS Model Forecasting Weekly Returns of the Tehran Stock Exchange Total Index Using the MIDAS Model 1 24 EN Journal Article 2026 06 22 <p>Forecasting capital market returns plays a pivotal role in the decision-making processes of investors, portfolio managers, and economic policymakers. Because economic variables are observed at different temporal frequencies, conventional temporal aggregation of data may result in the loss of valuable within-period information. The primary objective of this study is to evaluate the out-of-sample forecasting performance for weekly returns of the Tehran Stock Exchange (TSE) Total Index using a Mixed Data Sampling (MIDAS) regression model with a Beta polynomial weighting function. For this purpose, daily data on changes in the exchange rate, changes in the global gold price, changes in oil prices, daily index returns, and daily conditional volatilities extracted from a GARCH(1,1) model over the 2016–2025 period were employed. Model performance was evaluated using an out-of-sample forecasting-window approach by comparing forecast error measures, including the root mean squared error (RMSE) and mean absolute error (MAE), together with the Diebold–Mariano test of predictive accuracy. The empirical results indicate that changes in the exchange rate, the conditional volatilities of the exchange rate and gold price, and daily index returns are positively associated with weekly stock market returns, whereas the empirical evidence indicates a negative relationship between changes in the gold price and stock returns. Furthermore, incorporating high-frequency daily information and short-term volatility into the MIDAS specification resulted in a statistically significant reduction in out-of-sample forecast errors, as confirmed by the Diebold–Mariano test. The findings demonstrate the effectiveness of mixed-frequency models in extracting latent information from daily data to improve forecasting accuracy and risk management in the capital market.</p> https://bmfopen.com/index.php/bmfopen/article/download/633/467