技术择时信号20260109：A股仍维持乐观信号，看好小盘收益弹性

Quantitative Models and Construction Methods 1. Model Name: DTW Timing Model - Model Construction Idea: The DTW timing model is based on a similarity approach, analyzing the similarity between current index trends and historical trends. It selects historical segments with high similarity as references and calculates the weighted average future returns and standard deviations of these segments to generate trading signals [20][22]. - Model Construction Process: 1. Use the DTW (Dynamic Time Warping) distance algorithm instead of Euclidean distance to measure similarity, as DTW is better suited for time series problems by addressing sequence misalignment issues [22]. 2. Calculate the weighted average future returns and standard deviations of selected historical segments, where weights are the inverse of the DTW distance [20]. 3. Generate trading signals based on the average future returns and standard deviations [20]. 4. To address the "pathological matching" issue in traditional DTW algorithms, improved DTW algorithms with boundary constraints (e.g., Sakoe-Chiba and Itakura Parallelogram) are applied [24][26][28]. - Model Evaluation: The DTW timing model demonstrates stable excess returns in general market conditions but may underperform during periods of sudden macroeconomic policy changes [9]. 2. Model Name: Foreign Capital Timing Model - Model Construction Idea: This model leverages information embedded in the price movements of two offshore assets related to A-shares: FTSE China A50 Index Futures (Singapore market) and the Southbound A50 ETF (Hong Kong market) [30]. - Model Construction Process: 1. Construct two indicators using FTSE China A50 Index Futures: basis and price divergence [30]. 2. Construct a price divergence indicator using the Southbound A50 ETF [30]. 3. Combine the timing signals from the two assets to form the foreign capital timing signal [30]. - Model Evaluation: The model achieves strong performance, with annualized returns of 18.96% (long-short strategy) and 14.19% (long-only strategy) over the full sample period (2014-2024). It also exhibits a high win rate (close to 55%) and a profit-loss ratio exceeding 2.5 [13]. --- Model Backtesting Results 1. DTW Timing Model - Absolute Return: 35.52% since November 2022 [9] - Excess Return (relative to CSI 300): 8.60% [9] - Maximum Drawdown: 21.32% [9] 2. Foreign Capital Timing Model - Annualized Return (Long-Short Strategy): 18.96% (2014-2024) [13] - Annualized Return (Long-Only Strategy): 14.19% (2014-2024) [13] - Maximum Drawdown: 25.69% (Long-Short), 17.27% (Long-Only) [13] - Win Rate: Close to 55% [13] - Profit-Loss Ratio: Exceeds 2.5 [13] - Absolute Return (2024): 31.33% (Long-Only Strategy) [17] - Maximum Drawdown (2024): 8.23% [17] --- Quantitative Factors and Construction Methods 1. Factor Name: DTW Distance - Factor Construction Idea: DTW distance is used as a similarity measure for time series, addressing sequence misalignment issues that arise with traditional Euclidean distance [22]. - Factor Construction Process: 1. Compute the DTW distance between the current index trend and historical trends [22]. 2. Use the inverse of the DTW distance as weights to calculate the weighted average future returns and standard deviations of historical segments [20]. 3. Generate trading signals based on these weighted averages [20]. - Factor Evaluation: DTW distance is more effective for time series problems compared to Euclidean distance, as it resolves sequence misalignment and improves model performance [22]. 2. Factor Name: Basis and Price Divergence (Foreign Capital Timing Model) - Factor Construction Idea: These factors are derived from offshore assets to capture information about A-share market trends [30]. - Factor Construction Process: 1. Calculate the basis and price divergence indicators using FTSE China A50 Index Futures [30]. 2. Calculate the price divergence indicator using the Southbound A50 ETF [30]. 3. Combine these indicators to form the foreign capital timing signal [30]. - Factor Evaluation: These factors effectively capture offshore market signals and contribute to the strong performance of the foreign capital timing model [13]. --- Factor Backtesting Results 1. DTW Distance - Absolute Return: 35.52% since November 2022 [9] - Excess Return (relative to CSI 300): 8.60% [9] - Maximum Drawdown: 21.32% [9] 2. Basis and Price Divergence - Annualized Return (Long-Short Strategy): 18.96% (2014-2024) [13] - Annualized Return (Long-Only Strategy): 14.19% (2014-2024) [13] - Maximum Drawdown: 25.69% (Long-Short), 17.27% (Long-Only) [13] - Win Rate: Close to 55% [13] - Profit-Loss Ratio: Exceeds 2.5 [13] - Absolute Return (2024): 31.33% (Long-Only Strategy) [17] - Maximum Drawdown (2024): 8.23% [17]