In UNIST Financial Engineering Lab., we take quantitative approaches to financial planning of individuals and institutions. Most research topics would be categorized into the following three area.

• Investor analysis
  • We analyze investors' financial (e.g. income salary history, bank account balance history, credit card usage history) and non-financial data (e.g. occupation, family status, medical check-up history) to identify the key criteria for their financial planning (e.g. financial goals, risk-appetite, restrictions)
  • Relevant techniques: Data science; Statistics
• Financial market analysis
  • We study the dynamic behavior of various financial markets including equities, fixed-incomes, and commodities.
  • Relevant techniques: Econometrics; Machine learning
• Optimal investment decision
  • We derive customized and optimal financial plans based on investor data and market modeling. We are particularly interested in multi-stage decision making problems under uncertainties.
  • Relevant techniques: Optimization; Machine learning

Research Topics

We are working on many different research topics, but the followings are the projects that we are focusing on.

• Financial data analysis for individuals and households

  • A huge amount of financial data is generated by individuals and households, and they can be categorized into their asset, liability, income, consumption.
  • We can utilize this data to develop customized financial services for individuals and households to improve their financial well-being.
  • However, there are not many studies that analyze financial data of individuals and households.
  • Therefore, we aim to analyze various financial data of individuals and households and develop machine learning and/or optimization models to help them.
  • Examples of such data would be
    • Households’ asset, debt, income, and consumption (Korean Statistical Office)
    • Transactions in stock market (Shinhan Securities)
    • Track record of house mortgage (Korean Housing Finance Corporation)
    • Medical costs (National Health Insurance Service)
  • Related papers
    • Kim, Kyeongbin†; Hwang, Yoontae†; Lim, Dongcheol; Kim, Suhyeon; Lee, Junghye*; Lee, Yongjae** (2023) "Household Financial Health: A Machine Learning Approach for Data-Driven Diagnosis and Prescription," Quantitative Finance, conditionally accepted
    • Hwang, Yoontae; Lee, Yongjae**; Fabozzi, Frank J.* (2023) "Identifying Household Finance Heterogeneity via Deep Clustering," Annals of Operations Research, 325, 1255-1289
    • Park, Junpyo†; Hwang, Yoontae†; Kim, Jang Ho*; Lee, Yongjae**; Fabozzi, Frank J. (2023) “Heterogeneous Trading Behaviors of Individual Investors: A Deep Clustering Approach,” R&R (2nd round), Finance Research Letters

• Recommender systems for financial products

  • Recommender systems utilize a huge amount of user-item interactions to infer user preferences, and they have been key to success of various online retail services including YouTube, Amazon, and so on.
  • As we have seen from GameStop incident and Donghak Ant Movement, individual investors are now becoming more and more important in financial markets, and they do have their own preferences and perspectives.
  • Therefore, it would be useful to develop recommender systems for individual investors to help their investment based on their preferences.
  • However, investment cannot be done just based on preferences, because it is crucial to consider risk and return of investment.
  • Therefore, we develop recommender systems that address this unique challenge arising from the difference between financial products and other products.
  • Related papers
    • Choi, Minjoo†; Kim, Seonmi†; Kim, Yejin†; Lee, Youngbin†; Hong, Joohwan; Lee, Yongjae*** (2023) “An NFT Collectibles Recommender System with Content Features,” AI in Finance Bridge Program at AAAI-23
    • Kim, Seonmi; Lee, Youngbin; Kim, Yejin; Hong, Joohwan; Lee, Yongjae*** (2023) “NFTs to MARS: Multi-Attention Recommender System for NFTs,” submitted
    • Chung, Munki; Lee, Yongjae*; Kim, Woo Chang* (2023) “Mean-Variance Efficient Collaborative Filtering for Stock Recommendation,” submitted

• Analyzing time-series data and developing investment strategies using machine learning

  • As we have more and more data and computational power, data-driven modeling of financial time-series seems more promising.
  • Therefore, we try to utilize machine learning models to model and analyze financial time-series data and develop investment strategies.
  • Related papers
    • Kwon, Sohyeon; Park, Heehwan; Ahn, Dohyun; Lee, Yongjae* (2023) “Can GANs Properly Generate Financial Time Series?,” working paper
    • Hwang, Yoontae; Park, Junpyo; Lim, Dong-Young*; Lee, Yongjae** (2023) “Predicting Price Movement with Stop-loss Adjusted Labels,” R&R (2nd round), Finance Research Letters
    • Chung, Guhyuk; Chung, Munki; Lee, Yongjae*; Kim, Woo Chang* (2022) “Market Making under Order Stacking Framework: A Deep Reinforcement Learning Approach,” 3rd ACM International Conference on AI in Finance (ICAIF’22, acceptance rate: 42.0%)
    • Chung, Guhyuk; Lee, Yongjae*; Kim, Woo Chang* (2023) “Neural Marked Hawkes Process,” submitted

• Large-scale financial optimization

  • Financial planning is necessary for both institutions and individuals, and it often accounts for a multiple of decades.
  • Consequently, the problem becomes a large-scale optimization problem with large number of time periods and large number of decisions that need to be made.
  • Therefore, we try to develop large-scale financial optimization techniques to appropriately address such problems.
  • Relaxation, decomposition and function approximation techniques, such as semi-definite relaxation or stochastic dual dynamic programming (SDDP), are used.
  • Related papers
    • Kim, Min Jeong; Lee, Yongjae; Kim, Jang Ho; Kim, Woo Chang* (2016) "Sparse Tangent Portfolio Selection via Semi-Definite Relaxation," Operations Research Letters, 44(4), 540-543
    • Lee, Yongjae; Kim, Min Jeong; Kim, Jang Ho; Jang, Ju Ri; Kim, Woo Chang* (2020) “Sparse and Robust Portfolio Selection via Semi-Definite Relaxation,” Journal of the Operational Research Society, 71(5), 687-699
    • Kim, Woo Chang*; Kwon, Do-Gyun; Lee, Yongjae*; Kim, Jang Ho; Lin, Changle (2020) "Personalized Goal-Based Investing via Multi-Stage Stochastic Goal Programming," Quantitative Finance, 20(3), 515-526
    • Kim, Jang Ho; Lee, Yongjae; Kim, Woo Chang; Fabozzi, Frank J.* (2022) "Goal-Based Investing based on Multi-Stage Robust Portfolio Optimization," Annals of Operations Research, 313, 1141-1158
    • Lee, Jinkyu; Kwon, Do-Gyun; Lee, Yongjae*; Kim, Jang Ho; Kim, Woo Chang* (2023) “Large-scale Financial Planning via a Partially Observable Stochastic Dual Dynamic Programming Framework,” Quantitative Finance, accepted
    • Lee, Jinkyu; Bae, Sanghyeon; Kim, Woo Chang*; Lee, Yongjae* (2023) "Value Function Gradient Learning for Large-Scale Multistage Stochastic Programming Problems," European Journal of Operational Research, 308(1), 321-335
    • Bae, Hyunglip; Lee, Jinkyu; Kim, Woo Chang*; Lee, Yongjae* (2023) “Deep Value Function Networks for Large-scale Multistage Stochastic Programs,” Proceedings of The 26th International Conference on Artificial Intelligence and Statistics (AISTATS 2023), PMLR 206:11267-11287.
    • Bae, Sanghyeon; Lee, Yongjae*; Kim, Woo Chang* (2023) “Optimal Portfolio Choice of Couples with Tax Deferred Accounts and Survival Contingent Products,” R&R (2nd round), Quantitative Finance

• Uniformly distributed random portfolio (UDRP)

  • UDRP is a mathematical framework, developed by Kim and Lee (2016), that can represent the Sharpe ratio distribution of all possible portfolios.
  • We try to utilize this method to address various problems in investment domain.
  • The most natural application would be to evaluate fund managers' performance since UDRP provides the complete picture of risk-return outcomes.
  • Another way is to utilize UDRP's analytical expression to more thoroughly investigate puzzles in investment management.
  • Related papers
    • Kim, Woo Chang*; Lee, Yongjae (2016) "A Uniformly Distributed Random Portfolio," Quantitative Finance, 16(2), 297-307
    • Lee, Yongjae; Kwon, Do-Gyun; Kim, Woo Chang*; Fabozzi, Frank J. (2018) "An Alternative Approach for Portfolio Performance Evaluation: Enabling Fund Evaluation Relative to Peer Group via Malkiel’s Monkey," Applied Economics, 50(40), 4318-4327
    • Chung, Munki; Lee, Yongjae*; Kim, Jang Ho; Kim, Woo Chang; Fabozzi, Frank J.* (2022) “On the Effect of Mean, Variance, and Correlation on the Mean-Variance Framework,” Quantitative Finance, 22(10), 1893-1903