金融工程 证 券 研 究 报 告 中期：涨跌停模型中性。月历效应模型中性。 长期：长期动量模型看多。 综合：A 股综合兵器 V3 模型看空。A 股综合国证 2000 模型看空。 港股模型： 中期：成交额倒波幅模型看多。 【金工周报】（20250908-20250912） 指数择时多空互现，后市或中性震荡 ❖ 本周回顾 本周市场普遍上涨，上证指数单周上涨 1.52%，创业板指单周上涨 2.1%。 A 股模型： 短期：成交量模型所有宽基指数中性。低波动率模型中性。特征龙虎榜机构模 型看多。特征成交量模型看空。智能算法沪深 300 模型中性，智能算法中证 500 模型看空。 港股择时观点：上周我们认为看多港股指数，最终本周恒生指数涨跌幅为 3.82%。本周成交额倒波幅模型继续看多。因此我们认为，后市或继续看多恒 生指数。 下周推荐行业为：纺织服装、农林牧渔、电力设备及新能源、电力及公用事业、 综合金融。 ❖ 风险提示： 本周行业指数普遍上涨，除银行、综合金融、医药、消费者服务外所有行业均 上涨，涨幅前五的行业为：电子、房地产、农林牧渔、传媒、计算机。从资金 流向角度来说，除房地产、建筑、纺织服装外所有行业主力资金净 ...

Quantitative Models and Construction Methods Jump Imbalance Indicator - **Model Name**: Jump Imbalance Indicator - **Model Construction Idea**: Measures the difference in the strength of upward and downward jumps in stock prices[2] - **Model Construction Process**: - Formula: $$D_{i,t}^{N J}=\frac{\mathrm{No.of~Pjumps}_{i}^{d}\mathrm{\-~No.of~Njumps}_{i}^{d}}{\mathrm{No.of~Tjumps}_{i}^{d}}$$[14] - Parameters: - No.of Pjumps: Number of days with positive jumps in the past 20 trading days - No.of Njumps: Number of days with negative jumps in the past 20 trading days - No.of Tjumps: Number of days with jumps in the past 20 trading days[15] - **Model Evaluation**: Effective for timing the market but not outstanding[20] - **Model Testing Results**: - Annualized return: 6.23% - Sharpe ratio: 0.57 - Profit-loss ratio: 1.46 - Annualized excess return: 4.48% - Sharpe ratio (excess): 0.34[22] Implied Jump Imbalance Indicator - **Model Name**: Implied Jump Imbalance Indicator - **Model Construction Idea**: Reflects the jump information of stocks not affected by market jumps, potentially containing expectations of future performance or insider trading probability[23] - **Model Construction Process**: - Formula: $$D_{i,t}^{IJ}=\frac{\text{No.of Pumps}_{i}|\text{No Market Jump-No.of Numps}_{i}|\text{No Market Jump}}{\text{No.of Tumps}_{i}|\text{No Market Jump}}$$[23] - Parameters: - No.of Pjumps | No Market Jump: Number of days with positive jumps when the market index did not jump - No.of Njumps | No Market Jump: Number of days with negative jumps when the market index did not jump - No.of Tjumps | No Market Jump: Number of days with jumps when the market index did not jump[23] - **Model Evaluation**: Shows better performance compared to the Jump Imbalance Indicator[31] - **Model Testing Results**: - Annualized return: 9.93% - Sharpe ratio: 0.82 - Calmar ratio: 0.75 - Profit-loss ratio: 2.05 - Annualized excess return: 8.46% - Sharpe ratio (excess): 0.77 - Calmar ratio (excess): 1.12[34] Jump Imbalance Dispersion Indicator - **Model Name**: Jump Imbalance Dispersion Indicator - **Model Construction Idea**: Represents the dispersion of jump imbalance among stocks, indicating market sentiment divergence[39] - **Model Construction Process**: - Formula: $$\Delta J_{R_{Std}}$$[39] - Parameters: - Standard deviation of implied jump imbalance indicator among stocks[39] - **Model Evaluation**: Effective for timing the market[39] - **Model Testing Results**: - Annualized return: 9.41% - Sharpe ratio: 0.74 - Calmar ratio: 0.70 - Profit-loss ratio: 1.50 - Annualized excess return: 7.91% - Sharpe ratio (excess): 0.69 - Calmar ratio (excess): 0.72[42] Composite Indicator - **Model Name**: Composite Indicator - **Model Construction Idea**: Combines implied jump imbalance indicator and jump imbalance dispersion indicator for better market timing[40] - **Model Construction Process**: - Formula: $$\Delta J_{R} > 0 \text{ and } \Delta J_{R_{Std}} < 0$$[40] - Parameters: - Implied jump imbalance indicator - Jump imbalance dispersion indicator[40] - **Model Evaluation**: Shows significant improvement in market timing effectiveness[40] - **Model Testing Results**: - Annualized return: 16.5% - Sharpe ratio: 1.28 - Calmar ratio: 2.41 - Annualized excess return: 15.49% - Sharpe ratio (excess): 0.82 - Calmar ratio (excess): 0.88[45] Quantitative Factors and Construction Methods Jump Imbalance Factor - **Factor Name**: Jump Imbalance Factor - **Factor Construction Idea**: Measures the difference in the strength of upward and downward jumps in stock prices[2] - **Factor Construction Process**: - Formula: $$D_{i,t}^{N J}=\frac{\mathrm{No.of~Pjumps}_{i}^{d}\mathrm{\-~No.of~Njumps}_{i}^{d}}{\mathrm{No.of~Tjumps}_{i}^{d}}$$[14] - Parameters: - No.of Pjumps: Number of days with positive jumps in the past 20 trading days - No.of Njumps: Number of days with negative jumps in the past 20 trading days - No.of Tjumps: Number of days with jumps in the past 20 trading days[15] - **Factor Evaluation**: Effective for timing the market but not outstanding[20] - **Factor Testing Results**: - Annualized return: 6.23% - Sharpe ratio: 0.57 - Profit-loss ratio: 1.46 - Annualized excess return: 4.48% - Sharpe ratio (excess): 0.34[22] Implied Jump Imbalance Factor - **Factor Name**: Implied Jump Imbalance Factor - **Factor Construction Idea**: Reflects the jump information of stocks not affected by market jumps, potentially containing expectations of future performance or insider trading probability[23] - **Factor Construction Process**: - Formula: $$D_{i,t}^{IJ}=\frac{\text{No.of Pumps}_{i}|\text{No Market Jump-No.of Numps}_{i}|\text{No Market Jump}}{\text{No.of Tumps}_{i}|\text{No Market Jump}}$$[23] - Parameters: - No.of Pjumps | No Market Jump: Number of days with positive jumps when the market index did not jump - No.of Njumps | No Market Jump: Number of days with negative jumps when the market index did not jump - No.of Tjumps | No Market Jump: Number of days with jumps when the market index did not jump[23] - **Factor Evaluation**: Shows better performance compared to the Jump Imbalance Factor[31] - **Factor Testing Results**: - Annualized return: 9.93% - Sharpe ratio: 0.82 - Calmar ratio: 0.75 - Profit-loss ratio: 2.05 - Annualized excess return: 8.46% - Sharpe ratio (excess): 0.77 - Calmar ratio (excess): 1.12[34] Jump Imbalance Dispersion Factor - **Factor Name**: Jump Imbalance Dispersion Factor - **Factor Construction Idea**: Represents the dispersion of jump imbalance among stocks, indicating market sentiment divergence[39] - **Factor Construction Process**: - Formula: $$\Delta J_{R_{Std}}$$[39] - Parameters: - Standard deviation of implied jump imbalance indicator among stocks[39] - **Factor Evaluation**: Effective for timing the market[39] - **Factor Testing Results**: - Annualized return: 9.41% - Sharpe ratio: 0.74 - Calmar ratio: 0.70 - Profit-loss ratio: 1.50 - Annualized excess return: 7.91% - Sharpe ratio (excess): 0.69 - Calmar ratio (excess): 0.72[42] Composite Factor - **Factor Name**: Composite Factor - **Factor Construction Idea**: Combines implied jump imbalance factor and jump imbalance dispersion factor for better market timing[40] - **Factor Construction Process**: - Formula: $$\Delta J_{R} > 0 \text{ and } \Delta J_{R_{Std}} < 0$$[40] - Parameters: - Implied jump imbalance factor - Jump imbalance dispersion factor[40] - **Factor Evaluation**: Shows significant improvement in market timing effectiveness[40] - **Factor Testing Results**: - Annualized return: 16.5% - Sharpe ratio: 1.28 - Calmar ratio: 2.41 - Annualized excess return: 15.49% - Sharpe ratio (excess): 0.82 - Calmar ratio (excess): 0.88[45] Factor Backtesting Results Jump Imbalance Factor - **Annualized return**: 6.23% - **Sharpe ratio**: 0.57 - **Profit-loss ratio**: