金融工程 证 券 研 究 报 告 市场情绪监控周报（20260224-20260227） 深度学习因子 2 月超额 1.50%，本周热度变化最大 行业为钢铁、环保 深度学习因子跟踪 基于 DecompGRU 模型得分 TOP200 构建周度多头选股组合，组合样本外累计 绝对收益 84.37%，相对全指等权超额 43.11%；2 月组合绝对收益为 5.41%， 超额为 1.50%。 将个股得分聚合为 ETF 轮动组合，组合样本外累计绝对收益 53.40%，相对万 得主题 ETF 指数超额为 17.47%；2 月组合绝对收益为 9.51%，超额为 8.18%。 本周情绪因子跟踪 本周宽基热度变化方面：热度变化率最大的中证 1000，相比上周提高 3.63%， 最小的为中证 2000，相比上周降低 2.85%；宽基热度动量组合 26 年累计收益 为 5.6%。 本周申万行业热度变化方面，一级行业中热度变化率正向变化前 5 的一级行 业分别为钢铁、环保、公用事业、计算机、建筑材料，负向变化前 5 的一级行 业分别为食品饮料、商贸零售、纺织服饰、美容护理、社会服务；申万二级行 业中，热度正向变化率最大的 5 个行业是农 ...

- The DecompGRU model was used to construct a weekly long-only stock selection portfolio, holding the top 200 stocks with the highest integrated scores equally weighted The portfolio is rebalanced weekly based on the updated factor values from the previous Friday's closing prices Stocks with price limits or suspension are excluded, and transaction costs are not considered The benchmark is the CSI All Share Equal Weight Index[8][10] - The DecompGRU model's individual stock scores were aggregated to construct an ETF rotation portfolio The ETF pool is limited to industry and thematic ETFs, retaining only the ETF with the highest average daily trading volume over the past five days if multiple ETFs track the same index The portfolio is rebalanced weekly, holding 2-6 ETFs per period, with a benchmark of the Wind Thematic ETF Index[11][13] - A sentiment factor was constructed using user behavior data from Tonghuashun, aggregating stock-level heat indicators (browsing, watchlist, and click counts) normalized as a percentage of the total market and scaled by 10,000 This aggregated heat indicator serves as a proxy for "sentiment heat" at the broad-based index, industry, and concept levels[15][19][28] - A simple rotation strategy was built based on the weekly heat change rate (MA2) of broad-based indices, buying the index with the highest heat change rate on the last trading day of each week If the "Others" group has the highest change rate, the strategy remains in cash The strategy achieved an annualized return of 8.74% since 2017, with a maximum drawdown of 23.5%[21][24] - A concept-level sentiment strategy was constructed by selecting the top 5 concepts with the highest weekly heat change rates, excluding the bottom 20% of stocks by market capitalization within each concept From each concept, the top 10 stocks by total heat were equally weighted to form the "TOP" portfolio, while the bottom 10 stocks formed the "BOTTOM" portfolio The BOTTOM portfolio achieved an annualized return of 15.71% with a maximum drawdown of 28.89%[39][41][42] - The DecompGRU TOP200 portfolio achieved a cumulative absolute return of 74.91% and an excess return of 38.96% relative to the CSI All Share Equal Weight Index since its inception on March 31, 2025 The portfolio's maximum drawdown was 10.08%, with a weekly win rate of 68.18% and a monthly win rate of 100% In January 2026, the portfolio's absolute return was 8.99%, with an excess return of 0.98%[10] - The ETF rotation portfolio achieved a cumulative absolute return of 40.08% and an excess return of 5.93% relative to the Wind Thematic ETF Index since its inception on March 18, 2025 The portfolio's maximum drawdown was 7.82%, with a weekly win rate of 64.44% and a monthly win rate of 70% In January 2026, the portfolio's absolute return was 10.98%, with an excess return of 3.37%[13][14] - The broad-based index heat momentum strategy achieved a cumulative return of 6.6% in 2026[24] - The concept-level sentiment BOTTOM portfolio achieved a cumulative return of 3.7% in 2026[42]

Quantitative Models and Construction - **Model Name**: DecompGRU **Model Construction Idea**: The model improves information interaction between time-series and cross-sectional data by introducing two simple de-mean modules on the GRU baseline model[18] **Model Construction Process**: 1. The DecompGRU model architecture is based on GRU as the baseline 2. Two de-mean modules are added to enhance the interaction between time-series and cross-sectional data 3. The model is trained using IC and weighted MSE loss functions[18] **Model Evaluation**: The model demonstrates improved interaction between time-series and cross-sectional data, enhancing prediction accuracy[18] Model Backtesting Results - **DecompGRU TOP200 Portfolio**: - Cumulative absolute return: 41.11% - Excess return relative to WIND All A equal-weight index: 13.98% - Maximum drawdown: 10.08% - Weekly win rate: 64.52% - Monthly win rate: 100% - October absolute return: 1.78%, excess return: -0.07%[11] - **ETF Rotation Portfolio**: - Cumulative absolute return: 19.06% - Excess return relative to benchmark: -2.00% - Maximum drawdown: 7.82% - Weekly win rate: 62.50% - Monthly win rate: 57.14% - October absolute return: -2.04%, excess return: -1.18%[14][15] Quantitative Factors and Construction - **Factor Name**: Sentiment Heat Factor **Factor Construction Idea**: The factor aggregates stock-level sentiment heat metrics (e.g., browsing, self-selection, and clicks) to represent broader market sentiment[19] **Factor Construction Process**: 1. Individual stock sentiment heat is calculated as the sum of browsing, self-selection, and click counts 2. The sentiment heat is normalized by dividing by the total market sentiment on the same day and multiplying by 10,000 3. Aggregated sentiment heat is used as a proxy for market sentiment at the index, industry, and concept levels[19] **Factor Evaluation**: The factor effectively captures market sentiment and its impact on pricing errors[19] Factor Backtesting Results - **Broad-based Index Sentiment Heat Rotation Strategy**: - Annualized return since 2017: 8.74% - Maximum drawdown: 23.5% - 2025 portfolio return: 38.5% - Benchmark return: 32.9%[28] - **Concept Sentiment Heat BOTTOM Portfolio**: - Annualized return: 15.71% - Maximum drawdown: 28.89% - 2025 portfolio return: 42.1%[41][44]

Quantitative Models and Construction Methods - **Model Name**: DecompGRU **Model Construction Idea**: The model improves the GRU baseline by introducing two simple de-mean modules to enhance the interaction between temporal and cross-sectional information[14] **Model Construction Process**: 1. The DecompGRU model architecture is based on GRU with added de-mean modules for trend decomposition[14] 2. Two versions of the model are trained using different loss functions: IC and weighted MSE[14] 3. The IC-based model and MSE-based model are used to score stocks, and the top 200 stocks are selected for portfolio construction[8][14] **Evaluation**: The model effectively captures temporal and cross-sectional interactions, leading to improved stock selection performance[14] Model Backtesting Results - **DecompGRU Model**: - Cumulative absolute return: 24.54% - Excess return relative to WIND All A equal-weight index: 9.80% - Maximum drawdown: 10.08% - Weekly win rate: 72.22% - Monthly win rate: 100%[10] - **ETF Rotation Portfolio (Based on DecompGRU Scores)**: - Cumulative absolute return: 12.97% - Excess return relative to WIND ETF index: 8.65% - Maximum drawdown: 6.16% - Weekly win rate: 68.42% - Monthly win rate: 75%[12] Quantitative Factors and Construction Methods - **Factor Name**: Total Heat Indicator **Factor Construction Idea**: The indicator aggregates stock-level attention metrics (views, favorites, clicks) to represent market sentiment at broader levels (indices, industries, concepts)[17][18] **Factor Construction Process**: 1. Calculate the sum of views, favorites, and clicks for each stock[18] 2. Normalize the sum as a percentage of the total market activity on the same day[18] 3. Multiply the normalized value by 10,000 to derive the final indicator, with a range of [0, 10,000][18] **Evaluation**: The factor serves as a proxy for sentiment-driven mispricing, particularly effective at the stock level[18] Factor Backtesting Results - **Broad Index Heat Rotation Strategy**: - Annualized return since 2017: 8.74% - Maximum drawdown: 23.5% - 2025 YTD return: 18.8% - Benchmark return: 17.1%[27] - **Concept Heat BOTTOM Portfolio**: - Annualized return: 15.71% - Maximum drawdown: 28.89% - 2025 YTD return: 27%[44] Additional Observations - **Broad Index Heat Changes**: - Largest increase: CSI 500 (+10.21%) - Largest decrease: CSI 2000 (-6.02%)[27][29] - **Industry Heat Changes**: - Top 5 positive changes: Building Materials (+83.5%), Building Decoration, Social Services, Steel, Food & Beverage - Top 5 negative changes: Light Manufacturing (-32.5%), Textile & Apparel, Automotive, Real Estate, Utilities[38] - **Concept Heat Changes**: - Top 5 concepts: Dairy (+233.5%), Football (+194.9%), NMN (+115), Short Drama Games (+113.6%), Rent-Sale Rights (+109.6)[39][47][48]