Quantitative Models and Construction Methods 1. Model Name: GAN_GRU - **Model Construction Idea**: The GAN_GRU model combines Generative Adversarial Networks (GAN) for feature generation and Gated Recurrent Unit (GRU) for time-series feature encoding to construct a stock selection factor[4][13][14] - **Model Construction Process**: 1. **GAN Component**: - The generator (G) learns the real data distribution and generates realistic samples from random noise \( z \) (Gaussian or uniform distribution). The generator's loss function is: $$ L_{G} = -\mathbb{E}_{z\sim P_{z}(z)}[\log(D(G(z)))] $$ where \( D(G(z)) \) represents the discriminator's probability of classifying generated data as real[24][25][26] - The discriminator (D) distinguishes real data from generated data. Its loss function is: $$ L_{D} = -\mathbb{E}_{x\sim P_{data}(x)}[\log D(x)] - \mathbb{E}_{z\sim P_{z}(z)}[\log(1-D(G(z)))] $$ where \( D(x) \) is the probability of real data being classified as real, and \( D(G(z)) \) is the probability of generated data being classified as real[27][29][30] - GAN training alternates between optimizing \( G \) and \( D \) until convergence[30] 2. **GRU Component**: - Two GRU layers (GRU(128, 128)) are used to encode time-series features, followed by a Multi-Layer Perceptron (MLP) with layers (256, 64, 64) to predict returns. The final output \( pRet \) is used as the stock selection factor[22] 3. **Feature Input and Processing**: - Input features include 18 price-volume characteristics (e.g., closing price, turnover, etc.) sampled over the past 400 days, with a shape of \( 40 \times 18 \) (40 days of features)[18][19][37] - Features undergo outlier removal, standardization, and cross-sectional normalization[18] 4. **Training Details**: - Training-validation split: 80%-20% - Semi-annual rolling training (June 30 and December 31 each year) - Hyperparameters: batch size equals the number of stocks, Adam optimizer, learning rate \( 1e-4 \), IC loss function, early stopping (10 rounds), max training rounds (50)[18] 5. **Stock Selection**: - Stocks are filtered to exclude ST stocks and those listed for less than six months[18] - **Model Evaluation**: The GAN_GRU model effectively captures price-volume time-series features and demonstrates strong predictive power for stock returns[4][13][22] --- Model Backtesting Results 1. GAN_GRU Model - **IC Mean**: 0.1119*** (2019-2025)[4][41] - **ICIR (non-annualized)**: 0.89[42] - **Turnover Rate**: 0.83X[42] - **Recent IC**: 0.0331*** (December 2025)[4][41] - **1-Year IC Mean**: 0.0669***[4][41] - **Annualized Return**: 37.40%[42] - **Annualized Volatility**: 23.39%[42] - **IR**: 1.60[42] - **Maximum Drawdown**: 27.29%[42] - **Annualized Excess Return**: 22.42%[4][42] --- Quantitative Factors and Construction Methods 1. Factor Name: GAN_GRU Factor - **Factor Construction Idea**: The GAN_GRU factor is derived from the GAN_GRU model, leveraging GAN for price-volume feature generation and GRU for time-series encoding[4][13][14] - **Factor Construction Process**: - The GAN generator processes raw price-volume time-series features (\( Input\_Shape = 40 \times 18 \)) and outputs transformed features with the same shape (\( Input\_Shape = 40 \times 18 \))[37] - The GRU component encodes these features into a predictive factor for stock selection[22] - The factor undergoes industry and market capitalization neutralization and standardization[22] - **Factor Evaluation**: The GAN_GRU factor demonstrates robust performance across various industries and time periods, with significant IC values and excess returns[4][41] --- Factor Backtesting Results 1. GAN_GRU Factor - **IC Mean**: 0.1119*** (2019-2025)[4][41] - **ICIR (non-annualized)**: 0.89[42] - **Turnover Rate**: 0.83X[42] - **Recent IC**: 0.0331*** (December 2025)[4][41] - **1-Year IC Mean**: 0.0669***[4][41] - **Annualized Return**: 37.40%[42] - **Annualized Volatility**: 23.39%[42] - **IR**: 1.60[42] - **Maximum Drawdown**: 27.29%[42] - **Annualized Excess Return**: 22.42%[4][42] 2. Industry-Specific Performance - **Top 5 Industries by Recent IC (October 2025)**: - Social Services: 0.4243*** - Coal: 0.2643*** - Environmental Protection: 0.2262*** - Retail: 0.1888*** - Steel: 0.1812***[4][41][42] - **Top 5 Industries by 1-Year IC Mean**: - Social Services: 0.1303*** - Steel: 0.1154*** - Non-Bank Financials: 0.1157*** - Retail: 0.1067*** - Building Materials: 0.1017***[4][41][42] 3. Industry-Specific Excess Returns - **Top 5 Industries by December 2025 Excess Returns**: - Banking: 4.30% - Real Estate: 3.51% - Environmental Protection: 2.18% - Retail: 1.76% - Machinery: 1.71%[2][45] - **Top 5 Industries by 1-Year Average Excess Returns**: - Banking: 2.12% - Real Estate: 1.93% - Environmental Protection: 1.50% - Retail: 1.46% - Machinery: 1.23%[2][46]

Core Viewpoint - The article discusses the evolution and significance of reinforcement learning (RL) in China, highlighting key figures and their contributions to the field, particularly focusing on Wang Jun and his influence on the development of RL research and education in China [2][46]. Group 1: Historical Context and Development - Wang Jun's journey in AI began with information retrieval and recommendation systems, where he achieved significant academic recognition [4][8]. - His transition to reinforcement learning was influenced by his experiences in advertising, where he recognized the parallels between decision-making in advertising and RL principles [12][14]. - The establishment of the RL China community marked a pivotal moment in promoting RL research and education in China, addressing the lack of resources and formal education in the field [49][50]. Group 2: Contributions and Innovations - Wang Jun and his students have made substantial contributions to RL, including the development of SeqGAN and IRGAN, which integrate RL with generative adversarial networks for improved performance in various applications [23][24]. - The introduction of multi-agent systems in RL research has been a significant focus, with applications in complex environments such as advertising and gaming [27][28]. - The establishment of MediaGamma allowed for practical applications of RL in real-time advertising, showcasing the commercial viability of RL algorithms [17][18]. Group 3: Educational Initiatives and Community Building - The formation of RL China has facilitated knowledge sharing and collaboration among researchers and students, significantly enhancing the learning environment for RL in China [49][52]. - The publication of "Hands-On Reinforcement Learning" has provided accessible educational resources, bridging the gap between theory and practice for students [53]. - Wang Jun's mentorship has fostered a new generation of RL researchers, emphasizing the importance of exploration and innovation in academic pursuits [26][43]. Group 4: Future Directions and Challenges - The integration of RL with large models and embodied intelligence represents a promising frontier for future research, aiming to address the challenges of generalization across different tasks and environments [56][62]. - The ongoing exploration of RL applications in real-world scenarios, such as robotics and automated decision-making, highlights the potential for RL to impact various industries significantly [61][62]. - Despite setbacks in some projects, the commitment to advancing RL research and its applications remains strong among Wang Jun and his students, indicating a resilient and forward-looking approach to the field [56][62].