{"id":"arxiv/2610.31948","url":"https://arxiv.org/abs/2610.31948","tags":["cs.LG","math.ST"],"score":0,"title":"Sharp Bounds on the Winner's Bias under Independent and Nonnegatively Correlated Noise","domain":"cs.LG","source":"arxiv","authors":["Ivan Viakhirev","Kirill Borodin","Grach Mkrtchian"],"reviews":0,"subject":"cs.LG","summary":"Reporting the best of $M$ noisy, unbiased scores overstates the winner's value by up to $c\\sqrt{V}$, where $V$ is the total noise variance and the sharp constant $c$ depends on how the noise is dependent. It was known only at the two extremes, arbitrary dependence ($\\sqrt{(M-1)/M}$) and i.i.d. noise. We determine it for two classes in between: for independent, not identically distributed noise it is $\\sqrt{(M-1)/(2M)}$, and for pairwise nonnegatively correlated noise it is $(M-1)/M$; for uncorrelated noise this answers the maximum version of an open question about the expected range.…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-10-22","reactions":{}}