New York Session: One Session, Not One of Three
The New York session covers US trading hours, roughly 9:30am to 4pm Eastern for equities and longer for futures and currencies. For US stocks it is the only session rather than one of three, so its internal shape matters far more than any comparison with other sessions.
How it works
Equities trade 9:30am to 4pm Eastern. Index futures trade nearly around the clock but see most of their volume in those same hours, and currencies get their second big population then.
For a US stock, session analysis is not a comparison between three windows. There is one window. What matters is the shape inside it, and the premarket and after-hours periods either side, which are a different market with a fraction of the participation.
The shape of the day
Volume is heaviest at the open. Overnight orders execute, the opening gap gets traded, and everyone who formed a view outside hours acts at once.
The middle of the day is the thinnest part. Ranges narrow, spreads widen slightly, and a fixed execution cost becomes a much larger share of what a bar has to offer.
The close is an auction, not continuous trading. Index funds, benchmark-tracking accounts and anyone who must transact at the official close submit into it, and a very large share of the day’s volume prints in that single event at a price the auction determines.
Which is worth knowing for two reasons. The closing price on a daily chart is an auction price rather than the last trade of a continuous market — and the minutes before it can move on order imbalance rather than on anything a chart would explain.
Between the open and the close the profile is a shallow U, and it is remarkably stable across instruments and across years, because it reflects when people are at work rather than anything about a particular market. That stability is what makes it usable: you can plan around a curve that repeats, in a way you cannot plan around price.
All of that is a volume shape. The day has a reliable participation profile and no reliable directional profile, and conflating the two is the standard error in session material.
In practice
The first two hours overlap London and the rest of the session does not. So the session is not uniform: its first part has two populations trading and its second has one.
US economic releases cluster at 8:30 and 10:00 Eastern, and central bank statements at 2:00. Those are fixed, published times, which means a great deal of “session behaviour” is a response to a calendar you can look up in advance.
The opening gap is the session’s defining feature for stocks, because the market was closed for seventeen and a half hours and everything that happened arrives at once.
A stop triggering in the first minutes executes into the widest spreads of the day, which is the practical reason opening trades are expensive beyond what the chart shows.
The round trip is 2% of a median bar’s range on this site’s shared history — and on the smallest bar in the series it is 45%, which is the arithmetic that makes a quiet midday hour a bad time to trade small moves.
None of this structure exists above a daily chart. For a position held for weeks, the shape of a day is noise inside a bar.
Order book depth tracks the same profile. Deepest at the open and close, thinnest in the middle — which is where the tradeable size of any strategy is set.
Which produces a practical rule that has nothing to do with prediction: size the strategy to the thinnest hour you intend to trade in, not to the average. A method that works on paper at the open and cannot be filled at 1pm is two different methods wearing one name, and the difference only shows up in live fills.
What the New York session is not
It is not one of three equal sessions for stocks. It is the session; the others are premarket and after-hours.
It is not directionally biased. No hour of the day predicts which way price moves.
It is not uniform. The overlap hours and the afternoon are materially different markets.
And it is not the same for futures. Index futures trade nearly continuously, so their session structure is a volume curve rather than an open and a close.
When it fails
On a quiet day the structure is present and empty. Volume still peaks at the open and close, and price still goes nowhere. Session shape describes participation, and participation without disagreement produces no movement.
The second failure is trading the midday hours as though they were the open. Thin conditions and a fixed cost is the combination that quietly consumes accounts.
A third is the calendar confound. Results attributed to a time of day are often results attributed to a release schedule.
A fourth is the closing auction surprise. A position held into the close is exposed to an imbalance that is invisible until the last minutes.
And a fifth is treating premarket levels as ordinary structure. They formed on a fraction of the volume, and a level built on almost no participation is a level almost nobody is watching.
The original data
On this site’s shared 576-bar history the round-trip cost is 0.0098 price units — 2% of the median bar
range of 0.493 and 45% of the smallest bar of 0.022 — and it exceeds 10% of the bar’s range on 15 of the
576 bars, with ranges spanning 0.17 to 1.10 between the tenth and ninetieth percentiles. The figures are
in research/series-measurements.json, produced by site/measure_series.py.
Those 15 bars are the ones a session profile is really about. They are the quiet stretches where execution cost consumed more than a tenth of everything available, and on a real intraday chart they cluster in the middle of the day. Plot your instrument’s average bar range and average spread by half-hour, and the tradeable part of the session identifies itself — a chart you can build once from data you already have, and considerably more useful than a list of session names.
Related
Trading sessions is the parent page for the whole day. London session is the other half of the busiest overlap. And New York open covers the first minutes in detail.
Learning the shape of the day did more for my execution than any indicator. Not because the shape predicts anything, but because it told me when my costs were smallest and when I was paying for the privilege of trading a dead hour.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.