What Time of Day Does the Market Hit Its High? Nine Years of NQ Data

By 11am, 43 percent of NQ trading days have already printed their high of day. By 1pm it is 55 percent, by 3pm 73 percent, and by 3:30pm, 80 percent. That is the direct answer, measured across 2,213 regular-session trading days from 2017 through 2025.

The measurement is strict: a day counts at a given time only if its high is already in at that time and is never exceeded later in the session.

The full curve

When the extremes actually form

The distribution is heavily front-loaded. 43 percent of day highs print inside the 09:30 to 11:00 window. 27 percent print in the final hour. The middle of the session produces almost none. Day lows follow the same shape, slightly more front-loaded.

A rough out-of-sample check holds the pattern: in fresh 2026 sessions, 86 percent of days set at least one of their extremes in the first hour, against 78 percent across the nine-year sample.

What this means for afternoon reversal trading

If you are hunting the top at 2pm, you are not early. On six days out of ten, the high already exists and what you are watching is a retest or a lower high. That does not make afternoon shorts wrong, but it changes what they are: a bet on continuation of a structure that already formed, not a bet on catching the turn as it happens.

It also reframes midday. The lunch hours are not a quiet period to trade carefully through. They are a period in which the day's extreme structurally almost never forms. If your setup depends on catching an extreme, the careful response to midday is usually absence, not tighter stops.

Method notes

The count is purely descriptive: no model, no parameters, nothing tuned, which is also why it is reproducible by anyone with the same one-minute data. It describes NQ regular sessions specifically; the shape on ES is nearly identical and slightly flatter. A base rate like this is not an edge by itself. It is the denominator your intraday setups should be judged against, and most narratives about reversal timing are argued without one. For how we test whether any timing condition actually earns its place on real fills, the methodology is public in our case study.