Quantitative Models and Construction 1. Model Name: Heston Model - **Model Construction Idea**: The Heston model is used to calculate the implied volatility of near-month at-the-money options, serving as a market fear index. Implied volatility reflects market participants' expectations of future volatility [8]. - **Model Construction Process**: The Heston model is a stochastic volatility model where the variance of the asset price follows a mean-reverting square-root process. The model is defined by the following equations: $ dS_t = \mu S_t dt + \sqrt{v_t} S_t dW_t^1 $ $ dv_t = \kappa (\theta - v_t) dt + \sigma \sqrt{v_t} dW_t^2 $ Here: - \( S_t \): Asset price - \( v_t \): Variance - \( \mu \): Drift rate - \( \kappa \): Mean reversion speed - \( \theta \): Long-term variance - \( \sigma \): Volatility of variance - \( W_t^1, W_t^2 \): Correlated Wiener processes [8] 2. Model Name: Multi-Sector Timing Model (Scissor Difference Strategy) - **Model Construction Idea**: This model uses the difference in the number of bullish and bearish signals among sector constituents to construct a timing strategy. If no bullish or bearish signals are present, the scissor difference is set to zero. The model aims to outperform sector indices [16]. - **Model Construction Process**: - Count the number of bullish and bearish signals for each sector's constituent stocks daily. - Compute the scissor difference as the difference between bullish and bearish signals. - If both counts are zero, the scissor difference is set to zero. - Construct a timing strategy based on the scissor difference ratio [16]. - **Model Evaluation**: The model historically outperformed all sector indices, demonstrating excellent backtesting performance [16]. --- Model Backtesting Results 1. Heston Model - **Implied Volatility Results**: - SSE 50: 11.85% (down 0.88% WoW) - SSE 500: 14.35% (down 1.59% WoW) - CSI 1000: 18.06% (down 0.42% WoW) - CSI 300: 12.64% (down 0.73% WoW) [10] 2. Multi-Sector Timing Model - **Sector Outperformance**: The model outperformed all sector indices, achieving a 100% success rate in backtesting [16]. --- Quantitative Factors and Construction 1. Factor Name: Shape-Based Timing Signals - **Factor Construction Idea**: Shape-based signals are derived from historical K-line patterns, including bullish patterns (e.g., "Golden Needle Bottom," "Rocket Launch," "Manjianghong") and bearish patterns (e.g., "Hanging Line," "Paradise Line," "Dark Cloud Cover"). These patterns indicate potential price reversals [24]. - **Factor Construction Process**: - Identify specific K-line patterns based on predefined criteria. - Evaluate the historical performance of these patterns in predicting price movements. - Use the patterns to generate timing signals for individual stocks [24]. - **Factor Evaluation**: Bullish patterns like "Golden Needle Bottom" and "Rocket Launch" demonstrated strong positive predictive power [24]. --- Factor Backtesting Results 1. Shape-Based Timing Signals - **Signal Statistics**: - Positive signals: 2,699 occurrences, with an average future high-point success rate of 28.25% - Negative signals: 3,525 occurrences, with an average future low-point success rate of 71.88% [13] 2. Sector Timing Signals - **Bullish Sectors**: Home Appliances, Comprehensive, Communication, Textile & Apparel, Consumer Services, Transportation, Petrochemicals [19] 3. Stock-Specific Signals - **Consecutive Bullish Signals**: - 5-day signals: Stocks like Kailong Co. and Shipu Testing [21] - 4-day signals: Stocks like Jiangnan Chemical, Beijing-Shanghai High-Speed Railway, and Nandu Property [22][23] - **Special Bullish Patterns**: - Stocks like Retired Longyu ("Arrow on the String") and Suotong Development ("Manjianghong") [25][26] 4. Broker Gold Stock Signals - **Highlighted Stocks**: BYD, Feilihua, Wancheng Group, Sichuan Road & Bridge, Wolong Electric Drive, Lansheng Co., PetroChina, Dongpeng Beverage [29][33]