Medium-Term Forecasting Of Investment Portfolio Profitability

This study explores medium-term forecasting of investment portfolio profitability by analyzing the stock prices of six Uzbek joint-stock companies using time series models. The research compares classical statistical models such as ARIMA with nonlinear models like GARCH and LSTM to determine their accuracy in volatile market conditions. Over 848 ARIMA model combinations were tested, and the most optimal models were selected based on statistical indicators such as AIC, BIC, and significance of parameters. Findings revealed that combining ARIMA with GARCH models improves forecast precision due to the volatility observed in stock returns. The study also highlights that while residuals exhibit autocorrelation and non-normality, the models remain statistically robust for forecasting daily prices from August 2024 to December 2027. The research supports the need for hybrid approaches to better capture the dynamics of financial markets.

Hali tarjima qilinmagan