-
首页
-
- 首页
- WORKSHOP预告
数字金融Workshop：Building AI Models for Finance: Goal-Oriented Search
发布日期：2023-11-03 12:00 来源：
2023年秋季学期 北大数字金融Workshop 第七讲
Building AI Models for Finance: Goal-Oriented Search
时间：2023年11月03日周五 北京时间下午 2:00-3:30
Time: Friday, November 3 2023, 2-3:30 p.m. Beijing time
地点/Venue：
线下：北京大学国家发展研究院承泽园校区131教室
线上: 加入 Zoom 会议（会议号: 860 3656 7773 密码: 086222）
主讲人/Speaker: 丛林 Will Cong
主持人/Host：胡佳胤 Jiayin Hu
摘要/Abstract:
I discuss how the core theme in recent advances in AI can be adapted to finance
research. In Cong, Tang, Wang, and Zhang (2020), we build the first "large" model in finance to directly optimize the objectives of portfolio management via Transformer-based deep reinforcement learning. We develop multi-sequence, attention-based neural-network models tailored for the distinguishing features of financial big data, while allowing interactions with the market states and training without labels. Such AlphaPortfolio models yield stellar out-of-sample performances that are robust under various market conditions and economic restrictions. We further demonstrate AlphaPortfolio's flexibility to incorporate transaction costs, state interactions, and alternative objectives, before applying polynomial-feature-sensitivity and textual factor analyses to uncover key drivers of investment performance, including their rotation and nonlinearity.
We then extend this goal-oriented search using reinforcement learning to developing a data-driven-robust-control approach to corporate finance in Campello, Cong, and Zhou (2023). We treat managerial decision-making as a high-dimensional stochastic control problem with unknown market environment, non-linear impact, and dynamic
feedback. We first build a high-fidelity predictive environment module through supervised deep learning trained on corporate big data. The module can incorporate known stylized facts and results from reduced-form models and structural estimations, and generates more effective predictions and explanations of firm outcomes. We then apply model-based deep reinforcement learning to find the optimal managerial policy
for any given objective, and demonstrate significant room for improvement in modern enterprise decisions. Our approach also allows us to distinguish scenarios where theory and causal identifications are crucial from situations where predictive models trained on historical observations suffice.
Time-permitting, I will touch on panel trees developed in Cong, Feng, He, and He (2021) and Cong, Feng, He, and Li (2022) that offer an alternative and more interpretable way of searching for optimal solutions for economic questions such as test asset generation, pricing individual returns, and estimating sparse models on panel data with grouped heterogeneity.
主讲人介绍/Biography:
[Will Cong]
Lin William Cong is the Rudd Family Professor of Management and a tenured Professor of Finance at the Johnson Graduate School of Management at Cornell University, where he is the founding faculty director for the FinTech Initiative (https://www.linwilliamcong.com/fintech) and the founder of the Digital Economy and Financial Technology Lab (DEFT Lab, https://www.linwilliamcong.com/deft). He is also an Editor at the Management Science, an associate editor for multiple leading academic journals, a Research Associate at the National Bureau of Economic Research (NBER), a faculty scientist at the Initiative for Cryptocurrencies & Contracts (IC3), a lead founder of multiple international research forums (www.CBER-Forum.org and www.ABFR-Forum.org), and formerly a Kauffman Foundation Junior Faculty Fellow, a Poets & Quants World Best Business School Professor, a 2022 Top 10 Quant Professor, a finance professor and Ph.D. advisor at the University of Chicago Booth School of Business, George Shultz Scholar at the Stanford Institute for Economic Policy Research, and a doctoral fellow at the Stanford Institute for Innovation in Developing Economies. He has also advised leading investment and FinTech firms including Ansatz Capital, Ava Labs, Blackrock, Chainlink, DataYes, and Modular Asset management. He has also been invited to consult for or advise government and regulatory agencies such as the Asset Management Association of China, Bank of Canada, Department of Justice, the FBI, New York State Department of Financial Services, the New York State Office of Attorney General, and the U.S. Securities and Exchange Commission.
Professor Cong’s research spans financial economics, information economics, FinTech, digital economy, and entrepreneurship. He and his coauthors have pioneered the introduction of goal-oriented search and interpretable AI for finance, laid the foundations of tokenomics (covering categorization of tokens, cryptocurrency pricing, central bank digital currencies/payment systems, and optimal token monetary policy design), analyzed centralization issues and dynamic incentives in blockchains and DeFi, and developed data analytics for detecting market manipulation and better FinTech regulation, among others. He has won over 40 best paper prizes and research grants, including the Best Paper Prize at Management Science, and is a highly sought-after keynote speaker at various international conferences and forums. He has also been invited to speak or teach at hundreds of world-renowned universities, venture funds, investment and trading shops, and government agencies such as Alibaba, IMF, Monetary Authority of Singapore, and federal reserve banks.
国家发展研究院官方微信
Copyright© 1994-2012 北京大学 国家发展研究院 版权所有, 京ICP备05065075号-1
保留所有权利，不经允许请勿挪用
-
首页
-
数字金融Workshop：Building AI Models for Finance: Goal-Oriented Search
发布日期：2023-11-03 12:00 来源：
2023年秋季学期 北大数字金融Workshop 第七讲
Building AI Models for Finance: Goal-Oriented Search
时间：2023年11月03日周五 北京时间下午 2:00-3:30
Time: Friday, November 3 2023, 2-3:30 p.m. Beijing time
地点/Venue：
线下：北京大学国家发展研究院承泽园校区131教室
线上: 加入 Zoom 会议（会议号: 860 3656 7773 密码: 086222）
主讲人/Speaker: 丛林 Will Cong
主持人/Host：胡佳胤 Jiayin Hu
摘要/Abstract:
I discuss how the core theme in recent advances in AI can be adapted to finance
research. In Cong, Tang, Wang, and Zhang (2020), we build the first "large" model in finance to directly optimize the objectives of portfolio management via Transformer-based deep reinforcement learning. We develop multi-sequence, attention-based neural-network models tailored for the distinguishing features of financial big data, while allowing interactions with the market states and training without labels. Such AlphaPortfolio models yield stellar out-of-sample performances that are robust under various market conditions and economic restrictions. We further demonstrate AlphaPortfolio's flexibility to incorporate transaction costs, state interactions, and alternative objectives, before applying polynomial-feature-sensitivity and textual factor analyses to uncover key drivers of investment performance, including their rotation and nonlinearity.
We then extend this goal-oriented search using reinforcement learning to developing a data-driven-robust-control approach to corporate finance in Campello, Cong, and Zhou (2023). We treat managerial decision-making as a high-dimensional stochastic control problem with unknown market environment, non-linear impact, and dynamic
feedback. We first build a high-fidelity predictive environment module through supervised deep learning trained on corporate big data. The module can incorporate known stylized facts and results from reduced-form models and structural estimations, and generates more effective predictions and explanations of firm outcomes. We then apply model-based deep reinforcement learning to find the optimal managerial policy
for any given objective, and demonstrate significant room for improvement in modern enterprise decisions. Our approach also allows us to distinguish scenarios where theory and causal identifications are crucial from situations where predictive models trained on historical observations suffice.
Time-permitting, I will touch on panel trees developed in Cong, Feng, He, and He (2021) and Cong, Feng, He, and Li (2022) that offer an alternative and more interpretable way of searching for optimal solutions for economic questions such as test asset generation, pricing individual returns, and estimating sparse models on panel data with grouped heterogeneity.
主讲人介绍/Biography:
[Will Cong]
Lin William Cong is the Rudd Family Professor of Management and a tenured Professor of Finance at the Johnson Graduate School of Management at Cornell University, where he is the founding faculty director for the FinTech Initiative (https://www.linwilliamcong.com/fintech) and the founder of the Digital Economy and Financial Technology Lab (DEFT Lab, https://www.linwilliamcong.com/deft). He is also an Editor at the Management Science, an associate editor for multiple leading academic journals, a Research Associate at the National Bureau of Economic Research (NBER), a faculty scientist at the Initiative for Cryptocurrencies & Contracts (IC3), a lead founder of multiple international research forums (www.CBER-Forum.org and www.ABFR-Forum.org), and formerly a Kauffman Foundation Junior Faculty Fellow, a Poets & Quants World Best Business School Professor, a 2022 Top 10 Quant Professor, a finance professor and Ph.D. advisor at the University of Chicago Booth School of Business, George Shultz Scholar at the Stanford Institute for Economic Policy Research, and a doctoral fellow at the Stanford Institute for Innovation in Developing Economies. He has also advised leading investment and FinTech firms including Ansatz Capital, Ava Labs, Blackrock, Chainlink, DataYes, and Modular Asset management. He has also been invited to consult for or advise government and regulatory agencies such as the Asset Management Association of China, Bank of Canada, Department of Justice, the FBI, New York State Department of Financial Services, the New York State Office of Attorney General, and the U.S. Securities and Exchange Commission.
Professor Cong’s research spans financial economics, information economics, FinTech, digital economy, and entrepreneurship. He and his coauthors have pioneered the introduction of goal-oriented search and interpretable AI for finance, laid the foundations of tokenomics (covering categorization of tokens, cryptocurrency pricing, central bank digital currencies/payment systems, and optimal token monetary policy design), analyzed centralization issues and dynamic incentives in blockchains and DeFi, and developed data analytics for detecting market manipulation and better FinTech regulation, among others. He has won over 40 best paper prizes and research grants, including the Best Paper Prize at Management Science, and is a highly sought-after keynote speaker at various international conferences and forums. He has also been invited to speak or teach at hundreds of world-renowned universities, venture funds, investment and trading shops, and government agencies such as Alibaba, IMF, Monetary Authority of Singapore, and federal reserve banks.
访问学者申请
EMAIL
北京大学
ENGLISH
