If a bank can successfully predict the economic fundamentals of the country they operate in, they have a significant advantage when determining interest rate risk against their competitors. This in turn is advantageous when determining investment risk against their competitors. This paper explores using Reinforcement Learning (RL) as a method for predicting The United States' Gross Domestic Product (GDP) on a quarterly basis. Various RL algorithms are compared based on how accurately they predict the GDP output gap for the following quarter. This research was unable to accurately predict the GDP output gap on a quarterly basis, but further research could be done by including additional features and reward functions. Limitations for these findings are likely due to the small quantity of data on economics in the United States.
Swenson, Paul; Patel, Anish; Stroud, David; and Stacy, Jules
"Reinforcement Learning for Predicting the US GDP Output Gap,"
SMU Data Science Review: Vol. 6:
2, Article 5.
Available at: https://scholar.smu.edu/datasciencereview/vol6/iss2/5
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License