Shu Yu and Mulvey 2024 Dynamic Allocation
Yizhan Shu, Chenyu Yu and John Mulvey (Princeton ORFE), “Dynamic Asset Allocation with Asset-Specific Regime Forecasts” (arXiv:2406.09578, June 2024; published in Annals of Operations Research 346, 285-318). The paper extends the authors’ earlier 0/1 single-asset work (Shu Yu and Mulvey 2024) to a twelve-asset portfolio, using a Statistical Jump Model to identify per-asset regimes and an XGBoost classifier to forecast them, then feeding the forecasts into a Markowitz optimiser. Out-of-sample 2007-2023 with 5 bp one-way costs, the regime-aware portfolios beat their unconditional versions and a 60/40 benchmark — chiefly through roughly halved maximum drawdowns. It appears in this vault as the strongest favourable out-of-sample evidence for Regime-Based Asset Allocation, graded moderate because the return-forecasting correlation is only 2.43%, Sharpe gains are uneven, and the edge depends on turnover control.
Connections
- Regime-Based Asset Allocation — tests_strategy, source: https://arxiv.org/html/2406.09578v1
- Statistical Jump Model — proposes_model, source: https://arxiv.org/html/2406.09578v1
- Shu Yu and Mulvey 2024 — relates, source: https://arxiv.org/html/2406.09578v1
- John Mulvey — proposes_model, source: https://arxiv.org/html/2406.09578v1
- Transaction Costs and Slippage — includes_costs, source: https://arxiv.org/html/2406.09578v1
Sources
- Shu, Y., Yu, C. & Mulvey, J. M. (2024). “Dynamic Asset Allocation with Asset-Specific Regime Forecasts.” arXiv:2406.09578 / Annals of Operations Research 346, 285-318. https://arxiv.org/html/2406.09578v1