Portfolio Optimization and Resource Allocation
Sequential allocation tasks dynamically reallocate resources to a discrete set of targets. Portfolio optimization is an important application assigning financial resources to assets.
Research Focus
Our research is centered around reinforcement learning methods which a capable to handle policy function operating on a restricted polytope of continious assignments. Though we mostly evaluated our methods on the task of portfolio optimization, there are further applications like assigning computational resources to working tasks or funding to projects.
Publications
- David Winkel, Niklas Strauß, Maximilian Bernhard, Zongyue Li, Thomas Seidl, Matthias Schubert:Autoregressive Policy Optimization for Constrained Allocation Tasks. NeurIPS 2024
- David Winkel, Niklas Strauß, Matthias Schubert, Thomas Seidl: Simplex Decomposition for Portfolio Allocation Constraints in Reinforcement Learning, in Proceedings of the 26th European Conference on Artificial Intelligence (ECAI 2023),30.9-05.10 2023, Kraków, Poland
- David Winkel, Niklas Strauß, Matthias Schubert, Thomas Seidl: Constrained Portfolio Management using Action Space Decomposition for Reinforcement Learning, in Proceedings of 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2023), 25–28 May 2023, Osaka, Japan
- David Winkel, Niklas Strauß, Matthias Schubert, Thomas Seidl: Risk-Aware Reinforcement Learning for Multi-Period Portfolio Selection. ECML/PKDD (6) 2022: 185-200