Abstract Number:
1091
Submission Type:
Invited Paper Session
Participants:
Kiseop Lee (1), Jiwon Jung (1), Kiseop Lee (1), Yongdai Kim (2), Xiao Wang (1), Tim Leung (3), Sungkyu Jung (2), Hyoeun Lee (4), Jiwon Jung (1)
Institutions:
(1) Purdue University, West Lafayette, IN, (2) Seoul National University, Seoul, Korea, (3) University of Washington, Seattle, WA, (4) University of Illinois at Urbana Champaign, Champaign, IL
Chair:
Co-Organizer:
Session Organizer:
Speaker(s):
Hyoeun Lee
University of Illinois at Urbana Champaign
Session Description:
The recent surge in high-frequency market data, attributed to automated and algorithmic trading, has brought about the need for innovative methodologies to comprehend and decipher such complex datasets. Classical stochastic models often struggle to incorporate all new aspects of such data while maintaining their mathematical manageability. For instance, traditional stochastic models of financial markets often become too complicated to encompass and capture market frictions, price impacts, information effects, microstructure noises, bid and ask spreads and other factors that are now pivotal in high-frequency markets. Incorporating these previously ignored factors effectively within a sound model is challenging, and the most likely candidate models introduce difficulties in the analysis due to the inherent intricacy of financial markets. Consequently, data-driven approaches leveraging machine learning techniques have started garnering attention as alternatives.
In this session, we focus on data-driven methods to solve practical problems that traders face in today's financial markets. Our major tools include Reinforcement Learning, Deep Q learning, Double Q Learning, GAN, Transformer, and more. Given the nuanced nature of finance in the context of machine learning, unique hurdles arise. For instance, relying solely on historical data often falls short in accurately predicting the future dynamics of financial markets. This frequently calls for a fusion of human-machine interaction to enhance predictive capabilities . This becomes especially vital since the historical data lacks the social and political context of the current market, which can significantly influence price movements. In essence, even if a stock's movement appears similar, divergent future movements can occur due to differing social and political conditions, as well as human behaviors. Furthermore, the intense competition among traders renders even the most effective strategies short-lived. Therefore, traders must continually adjust their strategies to align with evolving market conditions.
Our featured six speakers, including two female scientists, with various backgrounds, will bring diverse expertise to address these issues through cutting-edge machine learning techniques. Prof. Yongdai Kim is an IMS fellow and the president of the Korean Artificial Intelligence Association. Prof. Xiao Wang is an ASA/IMS fellow and the co-director of the MS data science in finance program at Purdue University. Profs. Kim and Wang will discuss deep learning and generative models arising from financial markets. Prof. Sungkyu Jung is an expert in high-dimensional data analysis and Wasserstein quantile estimation, which are crucial in financial time series analysis. Prof. Tim Leung is the director of the University of Washington's MS Computational Finance and Risk Management program with extended experience in industry collaboration. Prof. Hyoeun Lee is a clinical assistant professor at UIUC, and she plays a major role in connecting students to financial industry jobs. Ms. Jiwon Jung is a Ph.D. candidate at Purdue University and has served as a data science consultant at AI consulting firm Vivity AI for a year. Ms. Jung is a young and promising female data scientist. She will present attention-based methods in financial forecasting.
This session has been designed to capture the attention of a diverse audience, including both academic researchers and practitioners.
Sponsors:
Business and Economic Statistics Section 1
Caucus of Industry Representatives 3
Korean International Statistical Society 2
Theme:
Statistics and Data Science: Informing Policy and Countering Misinformation
Yes
Applied
Yes
Estimated Audience Size
Small (<80)
I have read and understand that JSM participants must abide by the Participant Guidelines.
Yes
I understand and have communicated to my proposed speakers that JSM participants must register and pay the appropriate registration fee by June 1, 2024. The registration fee is nonrefundable.
I understand