Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.25821

Optimal Priors for Betting Tests with Bounded Payoffs: An Exact Efficiency Characterization

Francesco Orabona
View original source ↗Read source document ↗Export metadata ↓

Abstract

Testing by betting is a powerful framework that recasts hypothesis testing as a betting game. A popular approach uses an $F$-weighted portfolio with a predefined prior $F$. For suitable priors, this approach has recently been shown to achieve the optimal expected rejection time as the error probability goes to zero at a fixed alternative. Yet, the prior $F$ changes how the portfolio test behaves near the null. This motivates the question: Which prior should we use in testing-by-betting algorithms? We study bounded scalar payoffs and introduce an efficiency measure that allows the error probability and alternative to vary jointly. Under mild assumptions, we prove that efficiency equals a logarithmic decay index of the prior near zero. This gives a necessary and sufficient condition for optimality: Beta priors have infinite efficiency, whereas suitable Robbins-type priors attain efficiency 1. As an application, we consider testing a specified mean of a bounded random variable and describe an implementation of the portfolio algorithm with almost-linear computational complexity.

Explore related subjects

math.STcs.ITmath.ITstat.MEstat.TH

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Francesco Orabona. 2026-10-19. Optimal Priors for Betting Tests with Bounded Payoffs: An Exact Efficiency Characterization. https://arxiv.org/abs/2610.25821

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Detection Is Harder Than Estimation in Certain Regimes: Inference for Moment and Cumulant Tensors

We study estimation and detection of high-order cumulant tensors from $n$ i.i.d. observations of a $p$-dimensional random vector, with performance measured in the tensor spectral norm. Under sub-Gaussianity, we show that the minimax rate for estimating the order-$d$ cumulant tensor is $\sqrt{p/n}\wedge 1$. In contrast to covariance estimation, the natural plug-in estimator obtained by replacing population moments in the moment--cumulant formula with sample moments, which we call the sample cumulant tensor, is generally not rate-optimal for $d\ge 3$, and we construct an estimator that attains the minimax rate. On the computational side, we study testing whether the $d$-th order cumulant tensor vanishes after whitening. We provide evidence, under a model-specific low-degree conjecture, that detection is computationally hard when $n\ll p^{d/2}$. At the same time, we identify a regime in which an efficiently computable estimator achieves error smaller than the separation at which polynomial-time tests can reliably distinguish the null from the alternative under the same conjecture. This reveals an unusual reverse detection--estimation gap: computationally efficient detection can be harder than computationally efficient estimation. The underlying reason is that the relevant loss, tensor spectral norm, is itself NP-hard to compute, creating a new form of computational--statistical gap.

math.ST↗
arXiv2026

Debiased parametric inference for regime-switching jump diffusions with infinite variation jumps

We study high-frequency inference for an ergodic regime-switching jump diffusion with locally stable infinite-variation jumps. The state and the finite-state regime are observed on the same grid. The mesh satisfies \(h_n\to0\), while the horizon satisfies \(T_n=nh_n\to\infty\). We estimate structural drift and diffusion parameters in the presence of unknown regime-specific Lévy measures. For a common activity index \(1<β<8/5\), we remove two leading jump biases from each regime's truncated variation. We then combine the diffusion estimator with estimated signed tail means in an adaptive drift contrast. We prove joint asymptotic normality at rates \(\sqrt{T_n}\) and \(\sqrt n\), together with feasible studentization. The growing horizon requires uniform control of the debiasing remainders and random denominators over increasing regime exposures. It also produces an explicit Lévy-measure term in the drift covariance. A fixed-grid median stabilizes the two-step correction without changing its first-order limit. We study finite-sample behavior by simulation and apply the method to 127.5 hours of cattle sensor data. In the application, jaw-activity labels are recorded by a device separate from the sensor that measures the state process.

math.ST↗
arXiv2026

On the Accuracy of Approximate Leave-One-Out with Blockwise Nonsmooth Regularization in High-dimensions

Approximate leave-one-out (ALO) refers to methods that provide computationally efficient and statistically accurate approximations to leave-one-out risk estimation. These methods have recently attracted increased attention because of their connections to other important problems, including machine unlearning and data valuation. For empirical risk minimization, ALO methods often rely on a few Newton steps and therefore require smooth loss functions and regularizers. In high-dimensional problems, however, regularizers are frequently non-smooth. This has motivated the development of heuristic ALO methods for non-smooth regularization, whose empirical accuracy is often strong but whose theoretical justification remains limited. In this paper, we study the accuracy of a heuristic ALO method for a broad class of non-differentiable penalties that promote block sparsity in the proportional high-dimensional regime. Assuming uniformly bounded block sizes, we prove that the expected average absolute discrepancy between the exact leave-one-out loss and its ALO approximation is $O(\log^{3/2}(en)/\sqrt{n})$. Our proof technique differs substantially from existing approaches and relies on obtaining sharp bounds on the number of active groups whose status changes between the full-data and leave-one-out estimators. In addition to yielding stronger guarantees under weaker assumptions than prior work, our analysis provides new insight into the stability of sparse estimators under perturbations of the training data.

math.ST↗