Financial Portfolio Management with Deep Reinforcement Learning
T-Berger
Deep reinforcement has shown in the past, that it can successfully play complex games like the strategy board game go. Is it therefore also possible to analyze time series stock data to create a stock portfolio, which outperforms an etf-based benchmark?
This talk will show a top-down view of a basic implementation of a proximal policy optimization(PPO) reinforcement agent for portfolio optimization. As well, inform about regular problems of finance data in deep learning and show the advantages of deep reinforcement learning compared to time series forecasting for financial portfolio management. As a foundation for the talk, some basics of portfolio managements will also be shown and explained.
