Institution: Sungkyunkwan University
Advisor: Prof. Woocheol Choi, Prof. Jeonggyu Huh
Period: 2025
Overview
This project explored the combination of generative models and reinforcement learning for financial forecasting.
The framework integrated QuantGANs to generate synthetic financial paths and DDPG for portfolio policy learning, with the aim of forecasting future U.S. stock prices.
Contributions
- Built end-to-end modules for data preprocessing, QuantGANs training, synthetic path generation, and DDPG policy learning
- Applied the integrated framework to forecast future U.S. stock prices
GitHub
Slides
- QuantGANs — General introduction to QuantGANs
- TCN from QuantGANs — Detailed explanation of the TCN component within QuantGANs
- QuantGANs Analysis Seminar (2024) — Presented in the Analysis Seminar course at SKKU
- QuantGANs MIMIC 2024 — Seminar talk at the SKKU MIMIC | Video
- QuantGANs DDPG for Stock Prediction — Explanation of the integrated QuantGANs-DDPG implementation