<![CDATA[The foreign exchange (forex) market is characterized by high liquidity and rapid price fluctuations, which pose significant challenges for short-term price trend prediction using conventional statistical models. This study aims to develop a more reliable forecasting approach by addressing the limitations of standalone linear regression under volatile market conditions. To achieve this objective, a hybrid modelling framework is proposed that integrates Linear Regression with Mean Reversion principles and the Alligator technical indicator. Historical data OHLC data for three major currency pairs (AUDUSD, EURUSD, GBPUSD) were acquired from the Metatrader 5 platform, covering period from January to May 2025 with a 5-minute timeframe. The model performance was evaluated using monthly and weekly forecasting scenarios, supported by standard regression error metrics and back testing validation. The results show that compared to a baseline model using linear regression only, the proposed hybrid approach provides substantially improved predictive reliability in weekly forecasting scenarios, achieving average R2 values above 0.85 across all evaluated currency pairs. In addition, the hybrid framework facilitates clearer and more reliable buy and sell decision making by combining quantitative trend estimation with deviation base correction signals and trend structure confirmation. From a Computer Science perspective, this study contributes to predictive time series modelling and algorithmic trading research by demonstrating that hybrid integration of machine learning and technical indicators enhances model stability, interpretability, and decision support capability for short-term forex forecasting.]]>
