KIT - Chair of Financial Economics and Risk Management Teaching - Research and Thesis - Reinforcement Learning Projects - Simulating Stock Returns and Learning Optimal Portfolio Allocation via Imitation Learning (RL)

Modern Portfolio Theory (MPT), introduced by Harry Markowitz, is a fundamental framework for constructing optimal portfolios by balancing risk and return. It gives rise to the Efficient Frontier, which shows the portfolios offering maximum return for a given level of risk. However, real-world stock markets are complex, with varying conditions, non-linear asset relationships, and noisy returns, making static models less reliable.
Imitation learning (IL) provides a solution by training agents to mimic expert decisions, such as the Markowitz Efficient Frontier, through simulations. This project will simulate stock returns and use IL to teach an agent to replicate the optimal portfolio defined by Markowitz.