Data-driven Multiperiod Robust Mean-Variance Optimization

Paper: arXiv 2306.16681

Authors: Xin Hai, Gregoire Loeper, Kihun Nam

Abstract

We study robust mean-variance optimization in multiperiod portfolio selection by allowing the true probability measure to be inside a Wasserstein ball centered at the empirical probability measure. Given the confidence level, the radius of the Wasserstein ball is determined by the empirical data. The numerical simulations of the US stock market provide a promising result compared to other popular strategies.

Complexity vs Empirical Score

• Math Complexity: 7.5/10
• Empirical Rigor: 6.5/10
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper applies distributionally robust optimization using Wasserstein ambiguity sets to multiperiod portfolio selection, combining advanced measure-theoretic probability concepts with mean-variance framework. While supported by US stock market backtests, detailed code and replication assets are not fully disclosed.

Research Flowchart

  flowchart TD
    A[Research Goal: Multiperiod Portfolio Selection under Distribution Uncertainty] --> B[Data/Inputs: Empirical US Stock Market Data]
    B --> C[Methodology: Construct Wasserstein Ball around Empirical Measure]
    C --> D[Model Formulation: Data-Driven Distributionally Robust Mean-Variance Optimization]
    D --> E[Computational Process: Solve Multiperiod Robust Portfolio Strategy]
    E --> F[Findings: Superior Out-of-Sample Performance vs Standard Strategies]