---
title: Bayesian Optimization of CVaR with Unknown Hyperparameters
url: https://www.ml-quant.com/papers/arxiv/2610.18306/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-10-16
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: arXiv:2610.18306
source_url: https://arxiv.org/abs/2610.18306
featured: 2026-10-16
citations: unknown
topic: ML & AI Methods
---


# Bayesian Optimization of CVaR with Unknown Hyperparameters

HE-CV-UCB adaptively eliminates misspecified surrogate models and achieves no-regret Conditional-Value-at-Risk optimization when kernel hyperparameters are unknown.

- Source: https://arxiv.org/abs/2610.18306
- Identifier: arXiv:2610.18306
- Released: 2026-10-14
- First featured: Quant Letter No. 135 (2026-10-16): https://www.ml-quant.com/issues/2026-10-16/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: ML & AI Methods
- Authors: Valentin Margraf, Jonas Hanselle, Viktor Bengs, Serafima Lebedeva, Eyke Hüllermeier, Marcel Wever

## Abstract (arXiv, CC0)

We study Bayesian optimization of the Conditional Value-at-Risk (CVaR) of an expensive black-box function that depends on a decision and an environmental variable. Existing no-regret algorithms assume that the correct Gaussian process (GP) kernel and its hyperparameters are known. In practice, however, they must be estimated from limited data, often yielding a misspecified surrogate. This is especially damaging for CVaR because misspecification can distort the estimated tail-risk and thus lead to catastrophic decisions. We propose HE-CV-UCB, the first no-regret algorithm for CVaR Bayesian optimization with unknown GP hyperparameters. Given a finite candidate set containing a well-specified model, HE-CV-UCB adaptively eliminates those inconsistent with the observed data, actively queries where the survivors disagree, and returns a design that is robust across the surviving candidates. We prove that it attains cumulative regret matching the oracle rate of CV-UCB up to a factor depending only on the candidate set's size. Across synthetic benchmarks and three real-world applications, HE-CV-UCB achieves competitive performance and frequently outperforms existing CVaR Bayesian optimization methods.

## Related

- [Tail Risk Alert Based on Conditional Autoregressive VaR by Regression Quantiles and Machine Learning Algorithms](https://www.ml-quant.com/papers/arxiv/2412.06193/): The research uses AI to study tail risk in US financial markets, revealing a significant spillover effect from the credit market to the stock market.
- [Meta-models for transfer learning in source localization](https://www.ml-quant.com/papers/arxiv/2305.08657/): A Bayesian multilevel approach is used to predict model hyperparameters in acoustic emission experiments, demonstrating its use in source localization.
- [$\mu$GUIDE: a framework for quantitative imaging via generalized uncertainty-driven inference using deep learning](https://www.ml-quant.com/papers/arxiv/2312.17293/): Bayesian Framework for Imaging: The research introduces a Bayesian framework, $\mu$GUIDE, for estimating tissue microstructure parameters from any biophysical model or MRI signal, providing a solution to the high computational and time cost of traditional Bayesian methods.
- [What Do Foundation Models for Market Microstructure Learn? A Controlled Study with Synthetic Limit Order Books](https://www.ml-quant.com/papers/arxiv/2610.20104/): Autoregressive transformers trained on limit-order-book message streams learn amortized Bayesian inference over market regimes, matching optimal prediction and enabling accurate parameter calibration.
- [An Automated Hyperparameter Tuning Approach For Optimizing Machine Learning Model Performance](https://www.ml-quant.com/papers/ssrn/5234657/): The study suggests a new automated method for fine-tuning machine learning models using optimization techniques like Bayesian Optimization, Genetic Algorithms, and Reinforcement Learning.
- [Machine learning in algorithmic investment strategies on global stock markets](https://www.ml-quant.com/papers/repec/eee-riibaf-v-66-y-2023-i-c-s0275531923001782/): Algorithmic investment strategies using machine learning models perform better than passive strategies, with Linear Support Vector Machine and Bayesian Generalized Linear Model being the most effective, research shows.