Improved Genetic Algorithm‐Optimized SVR for Robust Long‐Term Forecasting of Global Stock Indices for Investment Decisions

ABSTRACT

Accurate long‐term forecasts are essential for high‐net‐worth individuals, institutional investors, and traders, yet long‐term multi‐step price forecasting remains a formidable challenge due to the inherent uncertainty over the long term. The proposed improved genetic algorithm‐optimized support vector regression (IGA‐SVR) model is specifically designed for long‐term price prediction of global indices. The genetic algorithm optimizes the SVR hyperparameters by minimizing a weighted average Mean Absolute Percentage Error (MAPE), assigning 20% weight to the full training dataset and 80% to the latest month validation window. The performance of the IGA‐SVR model is evaluated against four baseline models: Long Short‐Term Memory (LSTM), N‐BEATS, Transformer, and the forward‐validating genetic algorithm optimized support vector regression (OGA‐SVR). Extensive testing was conducted on the five global indices: Nifty, Dow Jones Industrial Average (DJI), DAX Performance Index (DAX), Nikkei 225 (N225), and Shanghai Stock Exchange Composite Index (SSE) from 2021 to 2024 on daily price prediction up to a year. Overall, the IGA‐SVR model achieved lower MAPE than other models, outperforming Transformer by 19.4%, LSTM by 24.5%, N‐BEATS by 41.3%, and OGA‐SVR by 55.1% on 1‐year daily price forecasting of global indices. Further, IGA‐SVR achieved an execution time reduction of 80% compared to N‐BEATS and 86% compared to LSTM and Transformer, highlighting its computational efficiency. Unlike rolling‐forward validation approaches that suffer from recency bias or deep learning models requiring heavy computation, IGA‐SVR provides a better accuracy and computationally efficient alternative for long‐term financial time series forecasting.