Schedule

The workshop date is 27 November 2023. All times are in the local New York time zone (Eastern Time). 

  8:30 - 10:30 First session
10:30 - 10:50 Break
10:50 - 12:30 Second session

First Session

8:30 - 9:05     Keynote: Ajim Uddin (NJIT) - Networked Prediction: Corporate Bond Return Forecasts through Institutional Holding Networks

9:05 - 9:20    Learning to Learn Network Momentum, Xingyue Pu, Stephen Roberts, Xiaowen Dong, Stefan Zohren

9:25 - 9:40    Financial Missing Investor Sentiment Imputation and Asset Pricing with A Novel Spatial-Temporal Graph Neural Network, Chang Luo, Tiejun Ma, Mihai Cucuringu 

9:45 - 10:00      A Comparative Analysis of Fine-Tuned LLMs and Few-Shot Learning of LLMs for Financial Sentiment Analysis, Sorouralsadat Fatemi, Yuheng Hu

10:05 - 10:20    Toward A Multiple Dimensional Risk Factors and Temporal Dependencies Transformer Model for Stock Price Movement Prediction, Hao Zhou, Tiejun Ma, Felipe Sperb

Second session

10:50 - 11:25  Keynote: Pranesh Srinivasan (Google) - Patterns and Anti-Patterns for AI/ML in Pari-Mutuel Settings

11:25 - 11:40  Company Similarity using Large Language Models, Dimitris Vamvourellis, Dhagash Mehta, Mate Toth, Dhruv Desai, Stefano Pasquali, Snigdha Bhagat

11:40 - 11:55  Improving Startup Success with Text Analysis, Emily Gavrilenko, Foaad Khosmood, Mahdi Rastad, Sadra Amiri-Moghadam 

11:55 - 12:30  Keynote: Jundong Li (University of Virginia) - Trustworthy Graph Machine Learning for Financial Applications

Keynote speakers

• Jundong Li (University of Virginia)

  • Title: Trustworthy Graph Machine Learning for Financial Applications

  • Abstract: Graph machine learning (GML) models, such as graph neural networks, have proven to be highly effective in modeling graph-structured data and achieving remarkable predictive performance in various high-stake financial applications, including credit scoring, fraud detection, and risk assessment. However, concerns have been raised regarding the trustworthiness of GML models in decision making scenarios when fairness, transparency, and accountability are lacking. To address these concerns, I will present our recent work on empowering GML for trustworthy financial decision making by focusing on two essential key aspects: fairness and explanation. First, I will discuss how to improve the fairness of GML from a data debiasing perspective. In particular, I will show how to measure data biases regarding different modalities of graph data and how to mitigate the data biases in a model-agnostic manner that can benefit different GML models. Second, I will show that explanation, as an effective debugging tool, not only can help us understand how the decisions are made but also could serve as a useful tool to diagnose how biases and discrimination are introduced in GML. Toward this goal, I will present a post-hoc structural explanation framework that can understand the unfairness issues of GML.

• Pranesh Srinivasan (Google)

  • Title: Patterns and Anti-Patterns for AI/ML in Pari-Mutuel Settings

  • Abstract: Peri-mutuel settings pose interesting challenges in the applications of AI/ML due to their dynamic nature. In this talk we go into the considerations of common techniques including problem decomposition, NLP and networks from the perspective of a practitioner and how they can prevent overfitting. 

• Ajim Uddin (NJIT)

  • Title: Networked Prediction: Corporate Bond Return Forecasts through Institutional Holding Networks

  • Abstract: This paper innovatively transforms institutional bond holdings into a network and develops an advanced Temporal Bipartite Graph Neural Network (TBGNN) model to extract pricing information for corporate bonds. Results show that our model can explain approximately 90% variations in in-sample returns and achieves at least a fourfold improvement in out-of-sample forecasts, compared to conventional linear and nonlinear models. By examining different network structures, data sparsity, and subsamples by bond characteriscs, we further demonstrate the consistent outperformance of our model over existing bond pricing factors. Our finding emphasizes that integrating the network information improves bond return forecasting. The findings are both economically and statistically significant and sheds light on the importance of a network-based model in asset pricing.

Accepted papers

• Company Similarity using Large Language Models, Dimitris Vamvourellis, Dhagash Mehta, Mate Toth, Dhruv Desai, Stefano Pasquali, Snigdha Bhagat [in-person]

• Learning to Learn Network Momentum, Xingyue Pu, Stephen Roberts, Xiaowen Dong, Stefan Zohren

• A Comparative Analysis of Fine-Tuned LLMs and Few-Shot Learning of LLMs for Financial Sentiment Analysis, Sorouralsadat Fatemi, Yuheng Hu

• Improving Startup Success with Text Analysis, Emily Gavrilenko, Foaad Khosmood, Mahdi Rastad, Sadra Amiri-Moghadam 

• Financial Missing Investor Sentiment Imputation and Asset Pricing with A Novel Spatial-Temporal Graph Neural Network, Chang Luo, Tiejun Ma, Mihai Cucuringu 

• Toward A Multiple Dimensional Risk Factors and Temporal Dependencies Transformer Model for Stock Price Movement Prediction, Hao Zhou, Tiejun Ma, Felipe Sperb