Bayesian Confidence Recalibration and Research-Equilibrium Criticality: Temporal Support in Robust Portfolios
Abstract
Robust portfolio rules that reconstruct confidence sets after learning need not preserve the evaluator obtained by prior-by-prior Bayesian transport. In the Gaussian model, this discrepancy is summarized by natural-coordinate displacement: inherited transport preserves it whereas fresh reconstruction can replace it. We price evaluator replacement and trace the resulting optimized curvature through endogenous research. Optimized robust value represents protocol regret as a functional Bregman divergence, while a within-vintage rectangular Gaussian benchmark with constant absolute risk aversion (CARA) yields a stopped recalibration tax. In a versioned model-release economy, validated history propagates through a strictly causal network and a same-cycle share of current optimized marginal value feeds back into research supply. The capacity-constrained equilibrium reduces to a scalar equation with protocol-indexed gain gIP=λβR,I(WRP)''. Purely causal validation cannot create a same-cycle unit mode; provenance changes criticality through optimized curvature. For a scalar primitive supplier-score shock z in direction hI and financial outcome 𝒪, sensitivity factors as ω𝒪,h,I/(1−gIP). Conditional on a smooth equilibrium state and active cell, completion-time information sharply bounds this multiplier when all compatible timing laws are subcritical; no finite uniform bound exists when the timing set reaches the pole.
Related
- Bayesian Fixed-Confidence Best-Arm Identification with Abstention
- Risk-Averse Bayesian Games with an Unknown Type Distribution: Bayesian Learning, Equilibrium Analysis, and Finite-Sample Guarantees
- A Path-Dependent Agent-Based Microsimulation of Crime and Violence Reduction Policy Portfolios in Bolivia: Survey Calibration, Adaptive Emulator Ensembles, Global Sensitivity, Tempered MCMC, and Multi-Objective Policy Analysis
- When Stress Tests Miss the Risk: Statistical Scenario Analysis for Financial Portfolios
- Assessing the impact of climate change and rising temperatures on life insurance portfolios
BibTeX
@misc{yanc2026bayesian,
title = {{Bayesian Confidence Recalibration and Research-Equilibrium Criticality: Temporal Support in Robust Portfolios}},
author = {Han Yanç},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.03741},
eprint = {2609.03741},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.03741},
}
RIS
TY - UNPB
TI - Bayesian Confidence Recalibration and Research-Equilibrium Criticality: Temporal Support in Robust Portfolios
AU - Han Yanç
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2609.03741
UR - https://arxiv.org/abs/2609.03741
AB - Robust portfolio rules that reconstruct confidence sets after learning need not preserve the evaluator obtained by prior-by-prior Bayesian transport. In the Gaussian model, this discrepancy is summarized by natural-coordinate displacement: inherited transport preserves it whereas fresh reconstruction can replace it. We price evaluator replacement and trace the resulting optimized curvature through endogenous research. Optimized robust value represents protocol regret as a functional Bregman divergence, while a within-vintage rectangular Gaussian benchmark with constant absolute risk aversion (CARA) yields a stopped recalibration tax. In a versioned model-release economy, validated history propagates through a strictly causal network and a same-cycle share of current optimized marginal value feeds back into research supply. The capacity-constrained equilibrium reduces to a scalar equation with protocol-indexed gain g_I^P=λβ_R,I(W_R^P)''. Purely causal validation cannot create a same-cycle unit mode; provenance changes criticality through optimized curvature. For a scalar primitive supplier-score shock z in direction h_I and financial outcome 𝒪, sensitivity factors as ω_𝒪,h,I/(1−g_I^P). Conditional on a smooth equilibrium state and active cell, completion-time information sharply bounds this multiplier when all compatible timing laws are subcritical; no finite uniform bound exists when the timing set reaches the pole.
ER -