• Factor Mining: Identified over 100 fundamental factors in the Chinese stock market at weekly and monthly frequencies, with Sharpe ratios surpassing the single-factor benchmark of 1.2, and automated the factor generator and back testing processes using Python• Factor Combination: Utilized genetic programming with DEAP for the combination optimization of daily price and volume factors, improving the Sharpe ratio of combinations by 0.3 compared to the single-factor benchmark

Knowledge-Driven Financial Quantitative Investment Techniques• Extracted key stock data from 20,000+ financial reports and identified 5 significant relationships among 300 stocks in Shanghai and Shenzhen with strong correlation performance.• Developed and fine-tuned a Graph Attention Network (GAT) model using PyTorch, modifying the adjacency matrix quarterly to analyze stock relationships.• Achieved significant improvement by using a cooperation relationship matrix, increasing the Information Ratio (IR) from 0.73 to 1.49, outperforming the fully connected matrix benchmark.

Research on Live-streaming Sales in E-commerce• Proposed and constructed models on 5 dimensions (emotions, relationship marketing, consumer engagement, brand personality, and perceived value) to explore performance of live-streaming sales in E-commerce• Award: Outstanding Research Project in the 40th – 41st Period of Research and Practice in SJTU (top 10%)

• Collected and analyzed Chinese market data with a specific focus on electronics• Assisted in writing 6 interim reviews and independently completed the interim report review of Transnational Holdings usingWind