{"id":"arxiv/2602.19481","url":"https://arxiv.org/abs/2602.19481","tags":["math.ST"],"score":0,"title":"A Selection Premium Decomposition for the Expected Maximum of Random Walks","domain":"math.ST","source":"arxiv","authors":["Victor H. de la Pena","Fangyuan Lin","Victor K. de la Pena"],"reviews":0,"subject":"math.ST","summary":"When $K$ models are evaluated on the same validation set of size $n$, the selected winner's apparent performance is biased upward. Suppose $K$ models are evaluated on a shared sequence of i.i.d. observations $X_1,\\dots, X_n$, where model $k$ achieves response $f_k(X_i)$ with mean $μ_k = \\mathbb E[f_k(X)]$. Writing $Y_{i,k} = f_k(X_i)-μ_k$ for the centered increment and $S_{n,k} = \\sum_{i=1}^n Y_{i,k}$ for the centered cumulative score, the expected maximum satisfies $0\\le\\mathbb E\\bigl[\\max_k S_{n,k}\\bigr] = \\sum_{i=1}^n \\mathbb E\\bigl[\\varphi_K(S_{i-1})\\bigr]$ where $\\varphi_K(u) =…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-02-23","reactions":{}}