——金融工程量化月报20260302：市场情绪较高，量化选股组合超额收益显著-20260302

Quantitative Models and Construction Methods 1. Model Name: PB-ROE-50 Strategy - Model Construction Idea: The PB-ROE strategy aims to identify expectation gaps in the market and enhance portfolio returns by incorporating surprise factors such as SUE and ROE growth[25] - Model Construction Process: - Based on the PB-ROE pricing model derived by Wilcox (1984), an expectation gap stock pool is created - From this pool, 50 stocks are selected using factors like standardized unexpected earnings (SUE) and year-over-year ROE growth to construct the PB-ROE-50 portfolio[25] - Model Evaluation: The strategy demonstrates the ability to generate positive excess returns in certain stock pools, though performance varies across different market segments[25] 2. Model Name: Institutional Research Strategy - Model Construction Idea: This strategy leverages public and private institutional research data to extract alpha by analyzing the frequency of company visits and stock performance relative to benchmarks before the visits[31] - Model Construction Process: - Public and private institutional research data are collected - Stocks are selected based on the number of times a company is visited and its performance relative to benchmarks before the visits[31] - Model Evaluation: The strategy effectively captures excess returns, with both public and private institutional research strategies showing positive results[31] 3. Model Name: Duration Rotation Strategy - Model Construction Idea: Based on the Nelson-Siegel model, this strategy uses three factors (level, slope, and curvature) to describe the dynamic changes in the yield curve and predict these factors for bond duration rotation[43] - Model Construction Process: - The Nelson-Siegel model is used to construct the yield curve - The model predicts the level, slope, and curvature factors - Based on these predictions, signals are generated for bond duration rotation[43] - Model Evaluation: The model provides actionable signals for bond allocation, such as the recent signal to allocate to long-duration interest rate bonds[43] --- Model Backtesting Results 1. PB-ROE-50 Strategy - Excess Return (February 2026): - CSI 500: -0.48% - CSI 800: 1.21% - All-market: 0.70%[25][28] - Year-to-Date Excess Return (2026): - CSI 500: -0.92% - CSI 800: 6.74% - All-market: 6.22%[28] 2. Institutional Research Strategy - Excess Return (February 2026): - Public Research: 2.22% (relative to CSI 800) - Private Research: 4.79% (relative to CSI 800)[31][34] - Year-to-Date Excess Return (2026): - Public Research: 1.36% - Private Research: 11.63%[34] 3. Duration Rotation Strategy - Latest Signal: Signal value of 10, indicating allocation to long-duration interest rate bonds[43][46] --- Quantitative Factors and Construction Methods 1. Factor Name: Percentage of Rising Stocks Sentiment Indicator - Factor Construction Idea: Strong-performing stocks often exhibit a demonstration effect, and the number of stocks with positive returns can indicate market sentiment[12] - Factor Construction Process: - Calculate the proportion of stocks in the CSI 300 index with positive returns over the past N days - Formula: $ \text{Percentage of Rising Stocks} = \frac{\text{Number of CSI 300 stocks with returns > 0 in the past N days}}{\text{Total number of CSI 300 stocks}} $[12] - Factor Evaluation: The indicator effectively captures upward opportunities but struggles to avoid risks in declining markets[12] 2. Factor Name: Moving Average Sentiment Indicator - Factor Construction Idea: Uses the eight-moving-average system to assess the trend state of the CSI 300 index by assigning values based on the number of moving averages above or below the current price[19] - Factor Construction Process: - Calculate the eight moving averages (parameters: 8, 13, 21, 34, 55, 89, 144, 233) - Count the number of moving averages above the current price - If the count exceeds 5, the sentiment is bullish[19] - Factor Evaluation: The indicator aligns well with the CSI 300 index's trend changes and provides clear signals for market sentiment[19] --- Factor Backtesting Results 1. Percentage of Rising Stocks Sentiment Indicator - Latest Value: Above 60% as of February 27, 2026, indicating high market sentiment[13] 2. Moving Average Sentiment Indicator - Latest Observation: The CSI 300 index is in a sentiment boom zone as of February 27, 2026[19]