Quantitative Models and Construction Methods - Model Name: Broad-based Index Rotation Strategy Model Construction Idea: The strategy is based on the weekly change rate of the "total heat" indicator for broad-based indices. It selects the index with the highest heat change rate for investment, while staying out of the market if the "Other" group has the highest rate[7][13][16] Model Construction Process: 1. Calculate the weekly heat change rate for each broad-based index (CSI 300, CSI 500, CSI 1000, CSI 2000, and "Other"). 2. Smooth the weekly change rate using a 2-period moving average (MA2). 3. On the last trading day of each week, invest in the index with the highest MA2 heat change rate. If the "Other" group has the highest rate, remain in cash. Model Evaluation: The strategy demonstrates a clear logic of capturing short-term market sentiment shifts based on heat changes[13][16] - Model Name: Concept Heat TOP and BOTTOM Portfolios Model Construction Idea: This model identifies the hottest concepts each week and constructs two portfolios based on the highest and lowest heat stocks within these concepts[30][32][34] Model Construction Process: 1. Select the top 5 concepts with the largest weekly heat change rates. 2. Exclude the bottom 20% of stocks in terms of market capitalization within the selected concepts. 3. Construct the "TOP" portfolio by equally weighting the top 10 stocks with the highest total heat within each concept. 4. Construct the "BOTTOM" portfolio by equally weighting the bottom 10 stocks with the lowest total heat within each concept. Model Evaluation: The model effectively captures behavioral biases in concept-driven markets, where low-heat stocks in hot concepts tend to generate excess returns over time[30][32][34] --- Model Backtesting Results - Broad-based Index Rotation Strategy: - Annualized Return: 8.74% - Maximum Drawdown: 23.5% - 2025 YTD Return: 14.7%[16] - Concept Heat BOTTOM Portfolio: - Annualized Return: 15.71% - Maximum Drawdown: 28.89% - 2025 YTD Return: 25.5%[34] --- Quantitative Factors and Construction Methods - Factor Name: Total Heat Indicator Factor Construction Idea: The factor aggregates the browsing, watchlist, and click counts of individual stocks, normalized as a percentage of the total market, and scaled by 10,000[7] Factor Construction Process: 1. Aggregate the browsing, watchlist, and click counts for each stock. 2. Normalize the aggregated value as a percentage of the total market. 3. Multiply the normalized value by 10,000 to obtain the total heat score. The range of the indicator is [0, 10,000][7] Factor Evaluation: The factor serves as a proxy for market sentiment and attention, effectively capturing behavioral patterns such as overpricing or underreaction at the stock level[7] --- Factor Backtesting Results - Total Heat Indicator: - No specific backtesting results provided for this factor in isolation
市场情绪监控周报(20250714-20250718):本周热度变化最大行业为房地产、公用事业-20250722