SCS(SH:601555) Soochow Securities·2025-05-06 04:16Quantitative Models and Construction Methods 1. Model Name: Single Technical Indicator Signal Model - Model Construction Idea: This model is based on price-volume data, utilizing various technical indicators to generate buy and sell signals. The goal is to adjust the position of an index to achieve excess returns[3][8] - Model Construction Process: - A total of 27 technical indicators were constructed and tested under specified backtesting conditions across three broad-based indices (CSI 300, CSI 500, CSI 1000) and 31 Shenwan first-level industry indices[8] - The indicators were designed based on the concept of price-volume "divergence" to capture potential trading opportunities[3][8] - Model Evaluation: The average annualized excess return of these indicators across 34 indices reached 3.75%, demonstrating their effectiveness in generating excess returns[3][8] 2. Model Name: Multi-Signal Combination Model - Model Construction Idea: This model combines multiple technical indicators through direct signal synthesis and rolling search methods to enhance performance and stability[3][8] - Model Construction Process: - Two strategies were developed: a 5-signal strategy and a 7-signal strategy - Signals were combined using correlation analysis to reduce redundancy and improve predictive power[3][8] - Model Evaluation: - The 5-signal strategy performed well on broad-based indices, achieving an annualized excess return of 11.27% on the CSI 1000 index[3][8] - The 7-signal strategy further refined the buy-sell distinction, improving performance in certain scenarios[3][8] 3. Model Name: Rolling Signal Combination Model - Model Construction Idea: This model uses rolling synthesis methods to combine signals, with two distinct approaches: post-merge buy-sell (Rolling Stable Strategy) and pre-merge buy-sell (Rolling Momentum Strategy)[3][8] - Model Construction Process: - Rolling Stable Strategy: Signals are merged first and then processed, resulting in more stable performance suitable for low-risk investors - Rolling Momentum Strategy: Signals are processed first and then merged, emphasizing momentum and reducing missed opportunities, suitable for high-risk investors[3][8] - Model Evaluation: - The Rolling Stable Strategy achieved an average annualized excess return of 3.99% with lower volatility - The Rolling Momentum Strategy demonstrated stronger momentum-following capabilities but with slightly higher volatility[3][8] --- Model Backtesting Results 1. Single Technical Indicator Signal Model - CSI 300: Annualized excess return of 3.01%[10] - CSI 500: Annualized excess return of 4.27%[10] - CSI 1000: Annualized excess return of 4.81%[10] 2. Multi-Signal Combination Model - 5-Signal Strategy: - CSI 300: Annualized excess return of 3.24%[10] - CSI 500: Annualized excess return of 1.61%[10] - CSI 1000: Annualized excess return of -4.20%[10] - 7-Signal Strategy: - CSI 300: Annualized excess return of 3.24%[10] - CSI 500: Annualized excess return of 4.25%[10] - CSI 1000: Annualized excess return of -1.76%[10] 3. Rolling Signal Combination Model - Rolling Stable Strategy: - CSI 300: Annualized excess return of 3.49%[14] - CSI 500: Annualized excess return of 4.25%[14] - CSI 1000: Annualized excess return of 5.11%[14] - Rolling Momentum Strategy: - CSI 300: Annualized excess return of 3.23%[14] - CSI 500: Annualized excess return of 1.90%[14] - CSI 1000: Annualized excess return of 0.00%[14]