Optimal stochastic control in multi-period portfolio optimization
Project Member(s): Hinz, J.
Funding or Partner Organisation: Zurich University of Applied Sciences
Start year: 2016
Summary: In this project, we investigate and examine novel methods for obtaining solutions to specific discrete-time optimal control problems. Our approach is based utilizes linear state dynamics and convexity assumptions commonly satisfied in practical applications. We extend this approach to a class of optimal switching problems under partial observation and we exploit specific model features to improve numerical performance. Further we adopt duality techniques to assess distance-to-optimality of our approximate solutions and apply these methods to address multi-period decision problems arising in the context of portfolio optimization.
Hinz, J 2016, 'Using Convex Switching Techniques for Partially Observable Decision Processes', IEEE Transactions on Automatic Control, vol. 61, no. 9, pp. 2727-2732.
View/Download from: Publisher's site
Keywords: Markov Decision, Approximate dynamic programming, Least Squares Monte Carlo
FOR Codes: Investment Services (excl. Superannuation), Financial Mathematics