开源量化评论（110）：构建跑赢中证500的分析师组合

Quantitative Models and Construction Methods - Model Name: Analyst 50 Portfolio Construction Idea: The portfolio is designed to outperform the CSI 500 index by leveraging the superior performance of newly recommended stocks and the effectiveness of the earnings surprise factor in stock selection[25][27][34] Construction Process: 1. Select newly recommended stocks from the monthly broker top 10 stock pool based on whether they are repeated from the previous month[25] 2. Calculate the earnings surprise factor (SUE) for each stock in the newly recommended stock pool, and select the top 50 stocks based on SUE values. Initial weights are assigned based on the number of broker recommendations[25] 3. Adjust the portfolio's industry weights proportionally to match the latest CSI 500 index industry weight distribution. Within each industry, stock weights are adjusted proportionally to finalize the Analyst 50 Portfolio[25] Formula: $SUE = \frac{R_t - E_t}{\sigma(R_t - E_t)}$ - $R_t$: Actual financial report value - $E_t$: Analyst forecast value - $\sigma(R_t - E_t)$: Standard deviation of the difference between actual and forecast values[22] Evaluation: The portfolio dynamically captures market trends and high-volatility stocks, demonstrating strong adaptability and aggressiveness[34] Quantitative Factors and Construction Methods - Factor Name: Earnings Surprise (SUE) Construction Idea: Measure the degree to which actual financial report values exceed analyst forecast values to identify stocks with potential for outperformance[22][23] Construction Process: 1. Use quarterly forecast and actual values to calculate the SUE factor[26] 2. Integrate earnings previews, quick reports, and regular reports, and forward-fill factor values during financial report gaps[23] Formula: $SUE = \frac{R_t - E_t}{\sigma(R_t - E_t)}$ - $R_t$: Actual financial report value - $E_t$: Analyst forecast value - $\sigma(R_t - E_t)$: Standard deviation of the difference between actual and forecast values[22] Evaluation: The factor shows strong differentiation in stock performance, especially in newly recommended stocks, where the information absorption is slower compared to repeated stocks[23] Model Backtesting Results - Analyst 50 Portfolio: - Annualized return: 14.9% - Annualized excess return over CSI 500: 15.6% - Maximum drawdown: 38.5% - Annualized volatility: 24.2% - Sharpe ratio: 0.58[27][29][20] Factor Backtesting Results - Earnings Surprise (SUE): - Newly recommended stocks: - Annualized return: 14.1% - Annualized volatility: 24.2% - Sharpe ratio: 0.58 - Maximum drawdown: 38.5% - Repeated stocks: - Annualized return: 8.8% - Annualized volatility: 23.3% - Sharpe ratio: 0.38 - Maximum drawdown: 45.0%[20][19][23]