BOCIC(SH:601696) Bank of China Securities·2025-07-04 15:09Quantitative Models and Construction Methods Model 1: High Prosperity Industry Rotation Strategy (S1) - Model Construction Idea: The model aims to select industries with upward profit expectations by tracking industry profitability using a multi-factor model based on analysts' consensus expectations[16] - Model Construction Process: - Construct three major types of factors based on the original value, slope, and curvature of profit expectations - Screen candidate factors with annualized excess return >3% - Use hierarchical clustering to classify candidate factors into 8 categories and select the highest excess return factor from each category for rank equal-weighted composite - Exclude overvalued industries and select the top 3 industries with the highest factor values weekly[16] - Model Evaluation: The model effectively captures industries with high profitability expectations[16] Model 2: Implicit Sentiment Momentum Tracking Strategy (S2) - Model Construction Idea: The strategy constructs a sentiment momentum model that runs ahead of earnings expectation data by capturing "unproven sentiment" in the market[19] - Model Construction Process: - Perform cross-sectional regression of industry daily returns on daily turnover rate changes to strip out "expected sentiment" - Calculate the residual as "unproven sentiment" - Construct half-month and 12-month momentum factors based on cumulative unproven sentiment factor net value - Rank and equal-weight composite the two momentum factors - Exclude overvalued industries and select the top 3 industries with the highest factor values weekly[20] - Model Evaluation: The model captures market sentiment ahead of earnings expectation data[19] Model 3: Macro Style Rotation Strategy (S3) - Model Construction Idea: The strategy predicts the long-short situation of four industry styles (high beta, high valuation, 12-month momentum, high volatility) based on current macro indicators and their correlation with the returns of these styles[22] - Model Construction Process: - Construct a fundamental indicator system from "economic growth," "inflation," "currency," "credit," and "market sentiment" - Calculate the exposure of each industry to the four styles and estimate the expected long-short returns of the style factors - Use a weak voting classifier to predict the long-short of the styles - Map the style predictions to industries and select the top 6 industries with the highest total scores monthly[23] - Model Evaluation: The model effectively integrates macro indicators with industry style predictions[22] Model 4: Long-term Reversal Strategy (S4) - Model Construction Idea: The strategy leverages the momentum effect within 2 years and the reversal effect beyond 3 years in industries[27] - Model Construction Process: - Construct a "1-year momentum" factor excluding the most recent month's returns - Construct a "3-year reversal" factor using the period from 3 years ago to 2 years ago - Construct a turnover factor using the turnover rate of free float market value - Rank and equal-weight composite the three factors - Select the top 5 industries with the highest factor values monthly[27] - Model Evaluation: The model captures long-term reversal and medium-term momentum effects in industries[27] Model 5: Fund Flow Industry Rotation Strategy (S5) - Model Construction Idea: The strategy constructs an industry rotation model based on "market main fund flow and strength" and "late trading fund flow and strength"[29] - Model Construction Process: - Construct an "institutional order trend strength factor" using the net buy amount of institutional orders - Construct a "late trading fund flow and strength factor" using the average daily inflow of late trading funds - Rank and equal-weight composite the two factors - Select the top 5 industries with the highest fund inflow strength monthly[30] - Model Evaluation: The model effectively captures the flow and strength of market funds[29] Model 6: Financial Report Factor Failure Reversal Strategy (S6) - Model Construction Idea: The strategy leverages the phenomenon of financial report factors performing poorly in recent years to construct an industry rotation model based on the mean reversion theory of factor effectiveness[34] - Model Construction Process: - Classify financial report factors into categories and screen for "long-term effective factors" with annualized excess return >5.5% - Identify "short-term failure factors" that underperform the industry equal-weight benchmark for 4 consecutive months - Composite the highest annualized excess return factors from each category - Select the top 5 industries with the highest factor values monthly[35] - Model Evaluation: The model captures the mean reversion of financial report factors[34] Model 7: Multi-factor Scoring Composite Strategy (S7) - Model Construction Idea: The strategy is a quarterly rebalancing strategy that composites factors from "momentum," "liquidity," "valuation," and "quality" dimensions[39] - Model Construction Process: - Exclude industries with a weight below 2% in the CSI 800 - Select 2 factors from each dimension and rank equal-weight composite - Select the top 5 industries with the highest factor values quarterly[40] - Model Evaluation: The model effectively integrates multiple factor dimensions[39] Model Backtest Results - S1: Annualized excess return -1.8% YTD[66] - S2: Annualized excess return 5.6% YTD[66] - S3: Annualized excess return 2.7% YTD[66] - S4: Annualized excess return 4.8% YTD[66] - S5: Annualized excess return -0.2% YTD[66] - S6: Annualized excess return 0.6% YTD[66] - S7: Annualized excess return 3.9% YTD[66] - Composite Strategy: Annualized excess return 2.0% YTD[66]