Paper: arXiv 2510.12725
Authors: Daniel Cunha Oliveira, Grover Guzman, Nick Firoozye
Abstract
Robust optimization provides a principled framework for decision-making under uncertainty, with broad applications in finance, engineering, and operations research. In portfolio optimization, uncertainty in expected returns and covariances demands methods that mitigate estimation error, parameter instability, and model misspecification. Traditional approaches, including parametric, bootstrap-based, and Bayesian methods, enhance stability by relying on confidence intervals or probabilistic priors but often impose restrictive assumptions. This study introduces a non-parametric bootstrap framework for robust optimization in financial decision-making. By resampling empirical data, the framework constructs flexible, data-driven confidence intervals without assuming specific distributional forms, thus capturing uncertainty in statistical estimates, model parameters, and utility functions. Treating utility as a random variable enables percentile-based optimization, naturally suited for risk-sensitive and worst-case decision-making. The approach aligns with recent advances in robust optimization, reinforcement learning, and risk-aware control, offering a unified perspective on robustness and generalization. Empirically, the framework mitigates overfitting and selection bias in trading strategy optimization and improves generalization in portfolio allocation. Results across portfolio and time-series momentum experiments demonstrate that the proposed method delivers smoother, more stable out-of-sample performance, offering a practical, distribution-free alternative to traditional robust optimization methods.
Complexity vs Empirical Score
- Math Complexity: 8.0/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper introduces a novel non-parametric bootstrap framework for robust optimization, involving complex statistical and optimization theory, and provides empirical results across portfolio and time-series momentum experiments, demonstrating out-of-sample performance improvement.
Research Flowchart
flowchart TD Start["Research Goal: Develop a robust framework for portfolio optimization under uncertainty without parametric assumptions"] --> Inputs["Data Inputs: Empirical financial time-series data<br>Uncertain parameters: Returns, Covariances, Utility"] Inputs --> Method["Methodology: Non-Parametric Bootstrap<br>Resample data to construct flexible confidence intervals<br>Treat utility as a random variable"] Method --> Process["Computational Process:<br>Percentile-based robust optimization<br>(Minimize worst-case outcome/Maximize lower percentile utility)"] Process --> Outcomes["Key Findings: <br>1. Mitigates overfitting & selection bias<br>2. Improves out-of-sample generalization<br>3. Delivers stable, smooth portfolio performance"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a non‑parametric bootstrap framework for robust optimization in finance.
Source sentence
This study introduces a non-Parametric bootstrap framework for robust optimization in financial decision-making.
Assumes 3
- Linear transaction costs are incurred on each trade. execution
Source sentence
tc represents a vector of transaction costs.
- A block bootstrap is applied to the return vector process to preserve dependence. statistical
Source sentence
To further mitigate overfitting, we propose applying the block bootstrap to the return vector process pt, denoted p∗t.
- Utility is modeled as a random variable with its own distribution. model
Source sentence
A key innovation is to treat utility as a random variable with an associated distribution, enabling optimization over distributional functionals—such as its α-th percentile.
Data
- ETF dataset market-prices commercial
Source sentence
Our empirical analysis relies on a panel of Exchange-Traded Funds (ETFs) with historical price data obtained from the WRDS database. The dataset covers multiple asset classes, including equities, fixed income, commodities, and volatility, providing a broad basis for the evaluation of portfolio and trading strategies. Th
- Futures dataset market-prices commercial
Source sentence
Our futures dataset is obtained from the Pinnacle database, which provides continuously linked contracts (CLC) built from individual futures series across major global exchanges.
Costs
Unclear
Finds 5
- Bootstrap‑based robust MVO (BUMVO 95) achieves the highest expected return (16.88%), strongest Sharpe (0.78), best Sortino (0.91) and lowest drawdowns (AvgDD –12.35%, MaxDD –40.20%). performance
Source sentence
BUMVO 95 achieves the highest expected return (16.88%), the strongest Sharpe ratio (0.78), the best Sortino ratio (0.91), and the lowest drawdowns (AvgDD –12.35%, MaxDD –40.20%).
- In the long‑short setting, BUMVO 75 provides the best balance with an expected return of 11.81% and the highest Sharpe (0.58) and Sortino (0.83) ratios. performance
Source sentence
BUMVO 75 provides the best overall balance, with an expected return of 11.81% and the highest Sharpe (0.58) and Sortino (0.83) ratios.
- Bootstrap‑based hyperparameter selection for time‑series momentum shows smaller and more stable out‑of‑sample declines (generally between –0.1 and –0.4) compared to ERM. performance
Source sentence
the bootstrap-based selections show smaller and more stable declines, generally between –0.1 and –0.4.
- Classical mean‑variance optimization is unstable out of sample, suffering from lower risk‑adjusted returns and larger drawdowns. negative
Source sentence
Classical mean–variance optimization is unstable out of sample, suffering from both lower risk‑adjusted returns and larger drawdowns.
- All bootstrap‑based methods systematically outperform classical MVO and ellipsoidal robust MVO across return, Sharpe and drawdown measures. performance
Source sentence
bootstrap‑based robust optimization (BUMVO) system‑atically outperforms both classical MVO and RPO across return, Sharpe, and drawdown mea‑sures.
Checked by others
Supported by
- An Analytic Solution for Asset Allocation with a Multivariate Laplace Distribution
Compare the findings
This paper: Classical mean‑variance optimization is unstable out of sample, suffering from lower risk‑adjusted returns and larger drawdowns.
That paper: Repeated mean‑variance optimization, correct for normally distributed returns, leads to overallocation to risk when returns are leptokurtotic.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (38)
- Time series momentum (2012) — link
- Portfolio Selection (1952) — link
- Distributionally Robust Neural Networks (2020) — link
- Statistics of Robust Optimization: A Generalized Empirical Likelihood Approach (2016) — link
- Robust Solutions of Optimization Problems Affected by Uncertain Probabilities (2013)
- Robust Supervised Learning (2005)
- Robust Portfolio: Optimization and Management (2012) — link
- Balanced Baskets: A New Approach to Trading and Hedging Risks (2012)
and 30 more
- Efficient Asset Allocation: A Practical Guide to Stock Portfolio Optimization and Asset Allocation (1998)
- Putting Markowitz Theory to Work (1981)
- On the Sensitivity of Mean-Variance Efficient Portfolios to Changes in Asset Means: Some Analytical and Computational Results (1991)
- Mean-Variance Revisited: Near-Optimal Portfolios and Sensitivity to Input Variations (1993)
- The Model Confidence Set (2011) — link
- A Reality Check for Data Snooping (2000) — link
- Improved Robust Portfolio Optimization (2017) — link
- The Probability of Backtest Overfitting (2016) — link
- Evaluating Trading Strategies (2014) — link
- Robust Portfolio Selection Problems (2003)
- Enhanced Portfolio Optimization (2021) — link
- Large Covariance Estimation by Hierarchical Clustering and Shrinkage (2022) — link
- Honey, I Shrunk the Sample Covariance Matrix (2004) — link
- Building Diversified Portfolios that Outperform Out of Sample (2016) — link
- Active Portfolio Management: A Quantitative Approach for Producing Superior Returns and Controlling Risk (1999)
- Resampling Methods for Dependent Data (2003)
- Bootstrapping Locally Stationary Processes (2014) — link
- Bootstrap methods for dependent data: A review (2011) — link
- 1 - Bootstrap Methods for Time Series (2012) — link
- Time series (re)sampling using Generative Adversarial Networks (2022) — link
- Robust Bayesian Portfolio Choices (2016) — link
- Robust control and model misspecification (2006) — link
- Maxmin expected utility with non-unique prior (1989) — link
- Percentile Criterion Optimization in Offline Reinforcement Learning (2023)
- Percentile Optimization for Markov Decision Processes with Parameter Uncertainty (2010)
- Bayesian Inference and the Parametric Bootstrap (2012)
- Robust Learning-Based Control via Bootstrapped Multiplicative Noise (2020) — link
- Robust Equity Portfolio Performance (2018) — link
- Incorporating Estimation Errors into Portfolio Selection: Robust Portfolio Construction (2016) — link
- Robust Asset Allocation (2004)
0 of 38 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.