Schedule

We present a (random) neural networks based algorithm which can solve nonlinear PDEs to price high-dimensional financial derivatives under default risk.

Ariel Neufeld is a Tenured Associate Professor in mathematics at the Nanyang Technological University in Singapore. He received his PhD in mathematics in May 2015 at ETH Zurich, where he spent half of his PhD at Columbia University in the City of New York. Prior to joining NTU he was a postdoctoral researcher at ETH Zurich. His research focuses on machine learning algorithms and their applications in finance and insurance, model uncertainty in financial markets and distributionally robust optimization, as well as stochastic analysis and stochastic optimal control. He was awarded in 2021 with the SIAM Activity Group on Financial Mathematics and Engineering Early Career Prize and recently with the Bruti-Liberati Visiting Fellowship Award.

A fundamental assumption in standard machine learning is that the training data follow the same probability distribution as the test data. However, in many real-world applications, this assumption is frequently violated due to factors such as evolving environments over time or sample selection bias driven by privacy concerns. This phenomenon, known as distribution shift, poses a significant challenge that must be addressed. In this talk, I will provide an overview of our research on tackling distribution shift, covering topics such as covariate shift, joint shift, sequential shift, and out-of-distribution adaptation.

Masashi Sugiyama received his Ph.D. in Computer Science from Tokyo Institute of Technology, Japan, in 2001. After serving as an assistant and associate professor at the same institute, he became a professor at the University of Tokyo in 2014. Since 2016, he has also served as the director of the RIKEN Center for Advanced Intelligence Project. His research interests include theories and algorithms of machine learning such as weakly supervised learning, noise-robust learning, and transfer learning. He was awarded the Japan Academy Medal in 2017 and the Commendation for Science and Technology by the Minister of Education, Culture, Sports, Science and Technology of Japan in 2022.

Generative models, while widely applied to simulate realistic effects in diverse domains, have seen limited exploration in virtual financial markets. Addressing this gap, we introduce the Large Market Model (LMM), an order-level generative foundation model designed for financial market simulation, akin to language modeling in the digital sphere. Our Financial Market Simulation engine, MarS, powered by LMM, sets a new standard by enabling realistic, interactive, and controllable order generation. Key insights reveal LMM's scalability with data and complexity, alongside MarS's robust realism and versatility. As a forecast tool, analysis platform, and training environment, MarS demonstrates transformative potential for financial applications. This paradigm shift opens new avenues for safe strategy testing and market behavior analysis, redefining financial simulations. Explore MarS's open-source implementation at https://github.com/microsoft/MarS/.

Weiqing Liu is currently a principal research manager in Machine Learning group in MSR Asia. He has been leading a team focusing on AI for Finance projects for several years. His current research is centered on the RD-Agent (https://github.com/microsoft/rd-agent) and MarS  (https://github.com/microsoft/mars) projects. He has authored tens of papers on top conferences. Before joining MSR Asia, he received his B.S. and Ph.D. degrees in Computer Science from the University and Science and Technology of China.

In this talk, I will share our observations on the potential applications of LLMs in professional financial scenarios. Centered around earnings call data, we explore novel generative tasks, including analyst report generation, generation-time filtering, and Q&A exercises. We will also highlight the challenges and difficulties in evaluation, as well as the importance of Human-AI Interaction in the LLM era. Finally, I will summarize our insights on the LLM era and share potential research directions.

Chung-Chi Chen is currently a researcher at the Artificial Intelligence Research Center, AIST, Japan. His scholarly pursuits revolve around the intricate realm of financial opinion mining and the nuanced understanding and generation of financial documents. He is the founder of ACL SIG-FinTech, and he has orchestrated the FinNLP/FinWeb workshop series within prestigious conferences such as IJCAI, WWW, EMNLP, and IJCNLP-AACL since 2019. He has guided the FinNum and FinArg shared task series on the NTCIR since 2018. He was also a presenter in the AACL-2020, EMNLP-2021, and ECAI-2024 tutorials. He served as Program Co-Chair of NTCIR-18, Senior Area Chair of ACL-2024, and PC member in many representative conferences. In academic competitions, he fortunately won the SIGIR Early Career Researcher Award (Excellence in Community Engagement), in addition to two Thesis Awards and Technology Innovation Award. Beyond academia, he has also ventured into the dynamic realm of FinTech. He earned one prize in a startup competition and four prizes in FinTech competitions. In addition to FinTech, he also has been honored with three prizes in LegalTech competitions.

Distributional reinforcement learning (RL) emerges as a powerful tool for modeling risk-sensitive sequential decisions, where leveraging distribution functions in place of scalar value functions has allowed for the flexible incorporation of risk measures. However, due to the inherent time inconsistency (TIC) in the use of numerous risk measures in sequential decision making, the nature of controls under distributional RL has remained a mystery. For its use in the risk-sensitive problems in mathematical finance, this paper seeks to fill the research gap by building on the cumulative prospect theory (CPT)-based analysis of human gambling behavior and the emergence of three policy classes under TIC: precommitment, equilibrium, and dynamically optimal. We focus on the prevailing quantile-based distributional RL (QDRL) for CPT risk measures. Our theoretical results extend some results from the risk-insensitive QDRL theory to CPT prediction, from which we derive the characterization of QDRL control as an approximate equilibrium of an intrapersonal game. We empirically demonstrate the efficacy of our CPT QDRL algorithm in approaching the equilibrium. Finally, by further exploring the economic interpretation of the three policy classes in their handling of TIC, we devise some metrics and instances relevant for driving interesting patterns of interactions between these policies, including when and how the equilibrium may be more desirable than the precommitment.

Patrick Pun is currently a tenured Associate Professor, Assistant Chair (MSc Programmes), and the Programme Director of Master of Science in Financial Technology at School of Physical and Mathematics Sciences, Nanyang Technological University, Singapore. Prior to NTU, Patrick obtained his Ph.D. in Statistics at the Chinese University of Hong Kong in 2016. His Ph.D. thesis won numerous awards, including Nicola Bruti Liberati Prize 2016 and the Young Scholars Thesis Award 2016. His research paper on high-dimensional portfolio selection won Best Student Research Paper (First Place) in INFORMS Financial Section in 2015. Patrick has strong research interests in Financial / Actuarial Mathematics, Big Data Analytics, and AI applications in Finance, as evidenced by his numerous top-tier publications in these fields.

The use of AI in hedge fund investment research is rapidly progressing from single-purpose models to compound AI systems, where multiple specialized agents work in concert to enhance both quantitative and fundamental research workflows. This session will explore real-world applications of generative AI, deep learning, and retrieval-augmented generation (RAG) in financial analysis — showcasing how these technologies drive faster, more accurate investment insights. We will examine key architectural considerations in designing multi-agent AI systems that can process large-scale, unstructured financial data, including earnings call transcripts, investor presentations, and news articles. Finally, we will discuss the practical challenges of embedding AI into hedge fund research workflows, covering optimization strategies, trade-offs in model complexity and performance, and best practices for ensuring reliability and trust in AI-driven research tools.

Chanyeol Choi is the Co-Founder and CEO of LinqAlpha, where he develops bespoke generative AI to streamline hedge fund investment research. He holds a Ph.D. in Electrical Engineering and Computer Science from MIT and was recognized in Forbes 30 Under 30 for his contributions to AI and computational research. His work spans over 30 publications with more than 4,000 citations, covering AI hardware, neuromorphic computing, and large-scale AI systems. At LinqAlpha, he leads the development of AI-native market intelligence and research management systems.

Joo Lee is the Co-Founder and CTO of Arrowpoint Investment Partners, an Asia-focused multi-strategy hedge fund backed by Blackstone, CPPIB, and Seviora (affiliate of Temasek). With over 15 years of experience building quantitative trading, risk management, and asset management systems in the financial industry, he previously served as the Founding CTO at Endowus, a pioneering digital wealth manager that, now in its fifth year, manages over US$7 billion in assets for more than 200,000 individuals, family offices, and institutions. Joo is an early adopter of artificial intelligence, using its potential to drive practical innovations in financial technology. He holds a Bachelor of Science in IT from the University of Technology Sydney.