Multiperiod bond portfolio optimization with transaction costs using a Markov Decision process

Battle Count: 50

Highly relevant for institutional fixed-income trading and bank treasury operations. It provides a rigorous framework for dynamic rebalancing under transaction costs, which is a core challenge in quantitative bond portfolio management. The use of MDPs and dynamic programming is standard in advanced quant finance.

Moderate to High. Requires constructing a time-inhomogeneous Markov chain from VAR simulations, handling state-space discretization and truncation, and implementing backward induction for the Bellman equation. The computational load is manageable for small state/action spaces but scales poorly without approximation techniques.

The paper provides detailed simulation parameters, including VAR parameters from Pericoli and Taboga (2016), discretization grids, and state space definitions. It states that no proprietary data was used and results are generated from simulations. However, specific code repositories are not explicitly linked in the text provided, though the methodology is fully described.