1Do one-minute financial markets see a firm's forthcoming accounting deterioration before daily markets do? We build a2strictly point-in-time panel of \nSampleFirms{} U.S. listed firms and \nSampleQuarters{} firm-quarters (2013--2026) from36.7 billion raw one-minute bars and quarterly statements with filing and release dates, and forecast an adverse4operating-income surprise twenty trading days before the earnings release. Holding the information set, the universe,5the horizon and the model capacity fixed, we add the \emph{same} families of realized measures, volume composition and6market co-movement computed from 30-minute, 5-minute and 1-minute bars to a benchmark that already contains point-in-time7fundamentals and daily OHLCV. Fundamentals explain most of the predictable variation, daily market data add a clear8increment (log loss \dligLBTwovsLBOne{} with gradient-boosted trees), and one-minute data add nothing: the paired9out-of-sample change in log loss is \dlogBFivevsBTwo{} [\lologBFivevsBTwo, \hilogBFivevsBTwo] with logistic regression10and \dligLBFivevsLBTwo{} [\loligLBFivevsLBTwo, \hiligLBFivevsLBTwo] with trees, intervals that exclude any gain at all with trees and any gain larger than a fifth of the daily11increment with logistic regression. The null survives five horizons, five accounting targets, a symmetric12volatility placebo, unseen firms, every refit year and every pre-specified subgroup; the small information content of13the incremental one-minute signal is concentrated in large, levered firms rather than in neglected ones, is already14impounded in prices at the cutoff, and has no economic value in factor-adjusted portfolio sorts. The accounting panel15is conditioned on firms listed in November 2026, a survivorship caveat we document rather than resolve. The invisible16balance sheet is visible in the daily close; the minute-by-minute tape adds noise, not foresight.17