Yue Wu

Attain to the broad and great while addressing the delicate and minute.

About Me

I’m a fourth year Ph.D. student from Academy of Mathematics and Systems Science (AMSS), Chinese Academy of Sciences, advised by Prof. Yongmiao Hong. I received my B.S. degree in Mathematics and Applied Mathematics from University of Chinese Academy of Sciences.

My research interest includes the intersection of AI, finance, and economics, with a particular focus on LLM Agent. I am always open to collaborations on related topics.

🔬 Research Portfolio: AI-Driven Financial & Macroeconomic Intelligence

(All as First Author)

I. Financial Market Prediction

Focusing on asset pricing, trading strategies, and market microstructure analysis.

1.Price Limit Information and Stock Market Prediction in China: A New Perspective Through Image Recognition (涨跌停信息与中国股市预测：图像识别的新视角)

  • 《管理科学学报》已录用，forthcoming（Journal of Management Sciences in China ;管理科学A级，FMS高质量T1期刊，北大核心，CSSCI）.
涨跌停制度作为中国A股市场的特色交易制度，对市场价格发现与投资者行为有着重要影响。为探索并利用这一信息，本文从市场情绪放大与价格反转效应两大理论机制出发，运用图像识别技术，基于2010年以来5353家A股成交量、价格及涨跌停收盘状态数据绘制趋势图，创新性地提出涨跌停状态线卷积神经网络(Line Convolutional Neural Network, LCNN)和涨跌停提示点卷积神经网络(Point Convolutional Neural Network, PCNN)两类方法。实证结果表明，相较于传统因子模型和结构化数据方法，融入涨跌停图像信息的模型在预测精度与投资组合收益方面具有显著优势，且优势在市场非理性程度较高（如高换手率、高波动率）的股票中优势更为突出。研究不仅验证了将关键市场制度信息“可视化”并用于量化预测的独特价值，也为中国特色市场机制下的涨跌停、融券制度优化提供了经验证据。研究结果有益于支持促进我国资本市场公平性和定价效率提升，进而推进中国金融市场长期高质量发展。

2.EvoTraders: An Evolutionary Multi-Agent System for Financial Trading

🌐 Online Demo · 💻 Code

EvoTraders is a multi-agent trading framework in which specialized analysts generate complementary signals, deliberate collectively, and adapt through performance-based updates and reflection-driven memory. Across seven real-world stock datasets, it consistently outperforms strong rule-based, reinforcement learning, and LLM-based baselines, achieving 24.15% annual return, 2.67 Sharpe ratio, and the lowest maximum drawdown.

II. Macroeconomic Forecasting

Focusing on news-driven narrative analysis, nowcasting, and systemic risk measurement.

3.When Large Language Model Meets Economic News: Advancing Narrative-Driven Macroeconomic Forecasting and Nowcasting

  • Best Student Presentation Award of 45th International Symposium on Forecasting（ISF, 2025）.
Proposes Target-LLM method that extracts multidimensional information from economic news through well-designed zero-shot prompts for macro prediction. Combines strengths of topic modeling and sentiment lexicon approaches, significantly outperforming traditional methods in forecasting US GDP, inflation, and unemployment. Constructed impact strength indicators provide more timely and accurate nowcasting performance.

4.Nowcasting and Decomposing Macroeconomic Risk via Business News and Large Language Model

  • The Australasian Econometric Conference of the Association of Econometricians Conference, 2025 (Sydney, Australia).
Introduces Risk-LLM framework that extracts high-frequency signals from economic news via zero-shot learning, embedded in the Growth-at-Risk model for macroeconomic risk measurement and decomposition. Substantially outperforms sentiment-dictionary benchmarks, especially during volatile periods, and more effectively identifies indirect effects in news. Fully replicable and adaptable to future LLM advances.

🎓 Education

Ph.D. (2022-2027， Advisor: Prof. Yongmiao Hong)
Academy of Mathematics and Systems Science, Chinese Academy of Sciences

B.S. in Mathematics and Applied Mathematics (2018-2022)
University of Chinese Academy of Sciences

📑 Talks

  • The Inaugural AE² Conference Australasian Econometric Conference of the Association of Econometricians (AE²), 2025 (Sydney, Australia).
  • 45th International Symposium on Forecasting, 2025 (Beijing, China).
  • Symposium on the Development of Economic Statistics in the New Era, 2025 (Xiamen, China).
  • Western Forum on Quantitative Economics, 2025 (Lanzhou, China).

🏆 Selected Awards

  • Best Student Presentation Award of 45th International Symposium on Forecasting（ISF）, 2025
  • Finalist, UBIQUANT Stock Trading Competition, UBIQUANT Quant Challenge Carnival, 2023
  • Academic Scholarship, University of Chinese Academy of Sciences (UCAS),2021
  • Academic Scholarship, University of Chinese Academy of Sciences (UCAS),2020
  • Academic Scholarship, University of Chinese Academy of Sciences (UCAS),2019
  • Merit Student, University of Chinese Academy of Sciences (UCAS), 2022
  • Merit Student, University of Chinese Academy of Sciences (UCAS), 2020

Flag Counter