Comparing Mixture, Box, and Wasserstein Ambiguity Sets in Distributionally Robust Asset Liability Management

Battle Count: 91

While the paper focuses on pension fund ALM rather than active trading, the DRO methodology and Wasserstein ambiguity sets are highly relevant to quantitative portfolio management. The framework for handling distributional uncertainty in asset allocation, the comparison of ambiguity set types, and the out-of-sample evaluation methodology are directly applicable to quantitative trading strategies. The Wasserstein DRO approach in particular has gained significant traction in quantitative finance for robust portfolio construction.

The paper involves multiple optimization formulations requiring LP duality transformations, SOCP reformulations, and Wasserstein distance computations. The mixture model is relatively simple (LP), the box model requires careful dual reformulation, and the Wasserstein model involves norm constraints and support set definitions. Scenario generation via Monte Carlo/GBM and market regime clustering add additional complexity. Solving the full multi-period ALM with all constraints requires commercial optimization solvers (e.g., Gurobi, CPLEX) for the SOCP formulations.

The paper provides detailed mathematical formulations and uses publicly available CPP data and market indexes. However, no code repository is mentioned. The Monte Carlo simulation parameters, specific index data processing, and k-mean clustering implementation details would need to be replicated. The regulatory constraints (X and Y sets) are specified but their exact implementation may vary.