Enhanced securities investment strategy using ISSA–SVM: a hybrid model combining adaptive moving average, support vector machine, and multi-strategy sparrow search algorithm for improved trend tracking and risk adjustment

Commodity Trading Advisor (CTA) strategies traditionally rely on fixed technical indicators for trend tracking, which often yield significant errors in volatile markets. This study proposes a novel hybrid strategy, ISSA–SVM, that combines Adaptive Moving Average (AMA), Support Vector Machine (SVM), and an Improved Sparrow Search Algorithm (ISSA) to enhance CTA model performance in securities investment. The strategy incorporates three key innovations: (1) an AMA algorithm that dynamically adjusts smoothing coefficients to market conditions, improving response speed and accuracy across different market environments. (2) SVM classification for trend prediction, leveraging its robust nonlinear classification capabilities. (3) a Multi-Strategy Sparrow Search Algorithm that optimizes multiple parameters and strategy combinations while avoiding local optima. Experimental results show that ISSA–SVM outperforms traditional CTA strategies, achieving an annualized return of 18%, a maximum drawdown of 12%, and high F1 scores across multiple market cycles. The strategy delivers stable positive returns with minimal risk exposure, demonstrating its strong market adaptability and superior risk-adjusted returns. During periods of high market volatility, ISSA–SVM performs significantly better, with a higher return-to-risk ratio than conventional models. These results establish ISSA–SVM as an effective enhancement to traditional CTA models, providing valuable insights for securities investment and presenting new opportunities for quantitative trading research.