{"id":"arxiv/2609.20550","url":"https://arxiv.org/abs/2609.20550","tags":["math.ST","q-fin.PM","q-fin.ST"],"score":0,"title":"Principal component error in high-dimensional factor models","domain":"math.ST","source":"arxiv","authors":["Alex Bernstein","Lisa R. Goldberg","Nicholas Gunther","Alec N. Kercheval","Tian Lan","Yian Lin","Dayi Yao"],"reviews":0,"subject":"math.ST","summary":"In a statistical factor model, principal components (or eigenvectors) of a sample covariance matrix serve as estimates of principal directions, the true drivers of co-movement of a collection of observed variables. We write the often substantial error in these estimates as a sum of two interpretable terms, which we show have almost sure asymptotic limits as the number of variables grows with sample size bounded. This scenario is commonplace in financial economics, genomics, machine learning and signal processing. Out-of-subspace error measures the distance from an estimate to the subspace…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-17","reactions":{}}