Id: arxiv/2603.09629 Url: https://arxiv.org/abs/2603.09629 Tags: math.ST Score: 0 Title: On the last time and the number of times an estimator is more than epsilon from its target value Domain: math.ST Source: arxiv Authors: Nils Lid Hjort, Grete Fenstad Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-03-10 Page: https://delphi.science/r/arxiv/2603.09629 Suppose $\widehatθ_n$ is a strongly consistent estimator for $θ_0$ in some i.i.d. situation. Let $N_\varepsilon$ and $Q_\varepsilon$ be respectively the last $n$ and the total number of $n$ for which $\widehatθ_n$ is at least $\varepsilon$ away from $θ_0$. The limit distributions for ${\varepsilon}^2 N_\varepsilon$ and ${\varepsilon}^2 Q_\varepsilon$ as $\varepsilon$ goes to zero are obtained under natural and weak conditions. The theory covers both parametric and nonparametric cases, multi-dimensional parameters, and general distance functions. Our results are of probabilistic interest,…