Dynamic Scenario-Based Portfolio Optimization for Multi-Asset ETFs

Paper: arXiv 2611.16540

Authors: Abigail A. Mensah, Ali M. Jaffri, Nicholas Appiah W. Brent Lindquist, Svetlozar T. Rachev, Frank J. Fabozzi

Abstract

Portfolio outcomes may depend as much on trading restrictions as on how returns are modeled. We compared historical moving-window returns with conditional scenarios from ARMA–GARCH models and a Student-$t$ copula for 30 U.S.-listed exchange-traded funds. Matched mean–variance and conditional value-at-risk portfolios were evaluated under long-only and limited long–short bounds over 946 sessions. Historical portfolios had higher observed reward–risk ratios in most comparisons, but no paired difference remained significant after multiple-testing adjustment. Conditional scenarios did not consistently reduce realized downside risk. We distinguished changes in target weights from trades required to rebalance holdings after market movements. Relaxing the target-adjustment limit changed exposures, turnover, and some returns more than varying scenario count or seed in the examined tests. These findings suggest that assessments of conditional return inputs should consider portfolio constraints and trading requirements.

Complexity vs Empirical Score

• Math Complexity: 7.5/10
• Empirical Rigor: 8.0/10
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper demonstrates strong empirical rigor through extensive backtesting, multiple statistical comparisons, and careful consideration of portfolio constraints. It also employs advanced mathematical models like ARMA-GARCH and Student-t copulas for scenario generation. The novelty lies in the detailed comparative analysis of historical vs. conditional scenarios under various trading restrictions, rather than proposing entirely new models.

Research Flowchart

  flowchart TD
    A[Research Goal: Assess Conditional Return Inputs' Impact on Portfolios] --> B{Key Methodologies: Mean-Variance & CVaR Optimization};
    B --> C[Data/Inputs: 30 U.S. ETFs, 946 Sessions, Historical vs. ARMA-GARCH Scenarios];
    C --> D{Computational Processes: Portfolio Rebalancing, Long-Only & Limited Long-Short Bounds};
    D --> E[Key Finding 1: Historical Portfolios Often Had Higher Reward-Risk Ratios];
    D --> F[Key Finding 2: Conditional Scenarios Did Not Consistently Reduce Realized Downside Risk];
    E & F --> G[Overarching Outcome: Portfolio Constraints & Trading Requirements Crucial for Assessing Conditional Return Inputs];

Cited References (32)

1. International asset allocation with regime shifts (2002)
2. On the sensitivity of mean-variance efficient portfolios to changes in asset means (1991)
3. Different approaches to risk estimation in portfolio theory (2004) — link
4. Generalized autoregressive conditional heteroskedasticity (1986)
5. The effect of errors in means, variances, and covariances on optimal portfolio choice (1993)
6. Empirical properties of asset returns: Stylized facts and statistical issues (2001)
7. Extreme value theory: An introduction (2006)
8. Optimal versus naive diversification: How inefficient is the 1/N portfolio strategy? (2009)

0 of 32 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2570 linked papers.