Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 23 min ago
PreprintCC BY 4.0arXiv

Neutral Is Not Free: Evaluating Downside Risk in Neutral Launches

  • Pablo Alcain
  • Jason Kang
  • Mia Garrard
  • Marine Veits
  • Houssam Nassif
  • Abbas Zaidi
Open at sourcePDFDOICite

Abstract

Evaluating "neutral launches" (e.g., infrastructure upgrades) using traditional confidence interval overlap is flawed: it is dangerously permissive with scarce data and excessively restrictive with abundant data. To resolve this, this paper introduces Expected Bayesian Loss (EBL), a continuous metric that quantifies both the probability and expected severity of metric degradation. Computable directly from standard frequentist estimates, EBL explicitly penalizes empirical noise and high-variance experiments. Validated against expert decisions, EBL provides experimentation platforms with a rigorous, tunable guardrail that aligns statistical safety with institutional risk appetite.

Related

  1. Downside Risk Optimization Under Asymmetric Laplace Distribution
    Zheng Liu, Wenhao Qian, Jingyi Wu +1
    MathematicsPeer-reviewedCC BY 4.0General mathematicsCrossref
  2. Gaussian Normalized Coordinates and Risk-Neutral CDF Deformations
    Jian Sun
    arXivPreprintCC BY 4.0Full textProbabilityarXiv
  3. GROMM: Generative Risk-neutral Option Market Models
    Magnus Wiese, Hans Buehler, Blanka Horvath
    arXivPreprintarXiv licenceFull textProbabilityarXiv
  4. Optimal Loss Allocation and Reflected Free Boundaries in Systemic Risk
    Yucheng Guo, Qinxin Yan
    arXivPreprintCC BY 4.0Full textOptimisation & controlProbabilityarXiv
  5. Exact Risk-Complexity Laws for Projective Boundaries in Scenario Optimization and Distribution-Free Certification
    Giuseppe C. Calafiore
    arXivPreprintarXiv licenceFull text1 citedOptimisation & controlarXiv

Cite

BibTeXRIS
BibTeX
@misc{alcain2026neutral,
  title = {{Neutral Is Not Free: Evaluating Downside Risk in Neutral Launches}},
  author = {Pablo Alcain and Jason Kang and Mia Garrard and Marine Veits and Houssam Nassif and Abbas Zaidi},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.10223},
  eprint = {2610.10223},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.10223},
  note = {CC BY 4.0},
}
RIS
TY  - UNPB
TI  - Neutral Is Not Free: Evaluating Downside Risk in Neutral Launches
AU  - Pablo Alcain
AU  - Jason Kang
AU  - Mia Garrard
AU  - Marine Veits
AU  - Houssam Nassif
AU  - Abbas Zaidi
PY  - 2026
DA  - 2026/10/07
PB  - arXiv
DO  - 10.48550/arxiv.2610.10223
UR  - https://arxiv.org/abs/2610.10223
AB  - Evaluating "neutral launches" (e.g., infrastructure upgrades) using traditional confidence interval overlap is flawed: it is dangerously permissive with scarce data and excessively restrictive with abundant data. To resolve this, this paper introduces Expected Bayesian Loss (EBL), a continuous metric that quantifies both the probability and expected severity of metric degradation. Computable directly from standard frequentist estimates, EBL explicitly penalizes empirical noise and high-variance experiments. Validated against expert decisions, EBL provides experimentation platforms with a rigorous, tunable guardrail that aligns statistical safety with institutional risk appetite.
ER  -