Fintech (Financial Technology) enables more people to create economic and social value at a much lower cost. However, there is currently no reliable fintech application for asset valuation, which is crucial for investment decisions. Existing automated trading applications, in general, adopt black box algorithms with no theoretical guidance, i.e., we have little idea why those algorithms perform well sometimes and under which conditions the algorithms would fail. In this project, we propose a new methodology for asset valuation by considering insights from classic economic theory and machine learning in data science. We first propose an indicator of asset valuation based on classic economic theory. Then we collect historical data for each argument in the simple indicator and utilize machine learning methods to assess the effectiveness of the simple indicator in explaining asset valuation and informing investment. Finally, we propose an automated trading strategy based on the simple indicator for further industry applications and test its performance in comparison to the conventional strategies. Junior and Senior Students who majored in Economics or Data Science are encouraged to apply.
Saad Lahrichi, Jiayi Li, Xinyu Tian, and Carlos Gustavo Salas Flores majoring in Data Science; Tianyu Wu (Student Project Lead) majoring in Mathematics and Computational Science, Duke Kunshan University
Luyao Zhang*, Tianyu Wu#, Saad Lahrichi#, Jiayi Li#, Carlos-Gustavo Salas-Flores#. “A Data Science Pipeline for Algorithmic Trading: A Comparative Study of Applications for Finance and Cryptoeconomics.” 2022 IEEE International Conference on Blockchain (Blockchain), Aug. 2022, 10.1109/blockchain55522.2022.00048.
Presented at the First International Symposium on Recent Advances of Blockchain Evolution: Architecture, Intelligence, Incentives, and Applications (BlockchianEvo 2022), Espoo, Finland, August 22-25, 2022, IEEE conference proceedings.