When Is the Best Time of Day to Trade Futures?
Two studies of the E-mini S&P 500 measure where volatility and order-book activity sit through the trading day, and neither of the two papers read for this post reports a return by time of day.

No hour is the best hour on the evidence read for this post. Two studies of the E-mini S&P 500 measure how far prices travel and how much traffic reaches the order book through the day, and both put their largest readings near the start and the end of US cash-market hours, with a sag in between — the U-shape both describe.
Activity is the environment a trade happens inside, and it is not the outcome of one. What follows is the session clock an account here runs on, the 24-hour volatility curve, whether it held still over fifteen years, what the order book does inside cash hours, and what the clock costs under a daily flat-time rule.
When is the best time of day to trade futures?
No best hour for the E-mini S&P 500 turns up in the evidence read for this post, because both studies read here measure the market's activity and not anyone's results. Between them they estimate how large price movements and order-book traffic are as the day progresses — one across the round-the-clock session, one inside US cash hours — and both put the largest readings near the open and the close.
What neither contains is a result for a trader. In the versions of the two papers read for this post, the word "profit" and the word "Sharpe" do not appear at all, and neither reports a return by time of day. The closest either comes is Takahashi's observation that ignoring the intraday variation can mis-state price impact and degrade trading and risk decisions.
The two documents carry different weight. Andersen, Su, Todorov and Zhang estimate the average intraday volatility curve over fifteen years of E-mini prices; the RePEc record puts the published version in the Journal of the American Statistical Association in 2024, volume 119, pages 1181–1191, while the copy read here is their working version dated 24 January 2023. Takahashi estimates the two-way relationship between returns and order-flow imbalance second by second and tracks the estimates through the day; that one is an arXiv preprint last revised 8 October 2025, and its arXiv record carries no journal reference, so nothing there records it as peer reviewed.
One limit applies throughout. Both papers are about the S&P 500 E-mini, so nothing in them says when gold, crude oil or Treasury futures are busiest. Treat the shape as a hypothesis to test on whatever you trade, in its own tick sizes and tick values.
When is the futures trading session actually open?
An account here runs on a trading day that opens at 6:00 PM ET and ends at 4:45 PM ET the next afternoon, when every position has to be flat; the 75 minutes in between are shut. The help center's trading hours article carries that rule and counts the trading day from one evening reopen to the next afternoon's flat time, so a Sunday-evening position belongs to Monday's trading day — which is also how winning days are tallied. A shortened holiday session moves the flat time to 12:45 PM ET.
Each paper describes the contract's exchange schedule as it stood over its own sample, and both describe the same two-layer arrangement: a long electronic session wrapped around a shorter core of US cash-market hours. Andersen and co-authors count 22.75 hours in a complete trading day and five such days a week, 17:00 to 16:00 Chicago time, Sunday through Friday, with trading halted 15:15–15:30 CT and maintenance from 16:00 to 17:00 CT. Takahashi's footnote puts regular trading at 8:30–15:15 CT under daylight saving, with a fifteen-minute break, electronic trading from 15:30 round to 8:30, and a half-hour maintenance shutdown daily.
Both describe a schedule in force during samples ending in 2020 and 2013, so check the current calendar before relying on a minute of either. The layering organises everything below: Takahashi narrows the empirical work to 8:30–15:00 CT, a window chosen to line up with the New York Stock Exchange's regular hours.
When is volatility highest during the futures day?
Volatility in the E-mini is highest during US hours, and within the business hours of each region the volatility study measures, it is highest at the two ends. Andersen, Su, Todorov and Zhang cut the 24-hour day at 2:00, 8:30 and 16:00 Chicago time, giving three blocks that cover active trading in Asia, Europe and America. What separates the blocks is scale. The American block's diurnal effect is the largest of the three and the Asian block's the smallest, and the paper describes the gap as a large multiple without ever putting a number on it. A further line in their chart marks the 15:15 CT halt, inside the American block.
The shape repeats inside all three blocks, which is the more useful half of the finding: volatility is elevated while a region's hours get going, sags through the middle, and picks up again toward their end.
The American block's last hour has a sequence of its own. The curve is still climbing into the 15:00 CT close of the US stock market, falls away quickly ahead of the 15:15 CT halt in futures trading, and is lower again across the 15:30–16:00 CT remnant of the day. Averaged across days, the open and the close sit significantly above the middle of the session, which is the established picture the paper restates — and an average is all that is: a given Tuesday can look like none of it.
The curve is estimated from 30-second transaction prices on the most active contract: 9,224,671 observations across 3,379 trading days, 30 November 2005 to 30 November 2020, with roll jumps adjusted away and returns measured across a gap in trading or a data fault dropped.
Has the intraday volatility pattern changed over time?
It has, and the volatility study puts that to a formal test. Andersen and co-authors split their fifteen years into three five-year subsamples — 991 trading days in 2005–2010, 1,140 in 2010–2015, 1,248 in 2015–2020 — and ask whether the volatility curve is the same in each pair.
The null that two curves are identical is rejected at the 1% level for 2005–2010 against each of the later periods; between the two later periods it reaches only the 10% level. Inside the American block the ranking is the same: 1% against the earliest period, 10% between the two later ones. The Asian and European blocks give weaker evidence, at the 5% and 10% levels depending on the pairing.
Where the curve moved has two parts. Inside the American block, volatility's weight shifted out of the middle of the session toward the close of US regular hours. Across the blocks, Asian hours took a larger relative share of the day's total volatility by the end of the sample, while the shape within those hours stayed roughly as it had been.
Two things qualify that. The hypothesis under test allows the curve to wobble from day to day in a stationary way, so rejecting it is evidence of a genuine change in the average shape. But the confidence bands around the 2015–2020 curve are wide across the entire day, where in the two earlier subsamples wide bands showed up mainly in the American region — a reason for caution about reading detail into the latest shape.
The practical reading is unglamorous: a rule of thumb about when the market goes quiet is a claim about a particular run of years, and across the fifteen years this study tested, the shape did not hold still.
When is the order book busiest, and when is it quiet?
Order-book traffic in the E-mini is heaviest at the two ends of the US cash session and lightest through the middle. Takahashi computes a set of activity measures on five-minute intervals from 8:30 to 15:00 CT and reports the same open-and-close-heavy U-shape in four of them: variability of mid-quote returns, variability of order-flow imbalance, number of events hitting the order book, and their average size.
The data is best-bid-and-offer files for the most active E-mini contract on each of 1,490 trading days from 2 January 2008 to 31 December 2013, events stamped to the second. The paper also fits a structural model to every fifteen-minute interval of each day — 38,740 fits, 37,029 of which met its rank condition — and tracks two parameters: how far an imbalance of a given size pushes price, and how strongly order flow answers a price move.
| Measure (E-mini S&P 500, 2008–2013) | Near the US cash open | Midday | Into the 15:00 CT cash close |
|---|---|---|---|
| Return volatility | High end of U; spike ~9:00 | Trough of U | Declines in the final five minutes |
| Order-book events and event size | High end of U; event count drops ~9:00 | Trough of U | Event size up at 14:55–15:00 |
| Price impact of an imbalance | Elevated 8:30–9:15; drops 9:15–9:30 | Peaks 13:00–13:15 | Falls into the close |
| Flow responsiveness to returns | Rises 9:00–9:30 | Lowest 12:45–13:00 | Surges 14:45–15:00 |
The top two rows are five-minute averages; the bottom two come from the fifteen-minute model, and their clock times are the paper's. For the activity measures the midday column is the shape the paper names — a trough between two busy ends — and not a reading it prints for the middle of the day. For the two model parameters it does give midday turning points, which are in the table.
How does a thin order book change price impact?
The less depth sitting on the book, the further one-sided order flow of a given size pushes price. Takahashi regresses the estimated price-impact parameter on market conditions and finds the reciprocal of depth entering strongly positive: impact grows as depth shrinks. A higher count of events and a larger average event size each come with lower impact, as does a wider spread. Conditions of that kind, plus time-of-day dummies, account for roughly 54% of the variation in the parameter.
Two moments show how differently those conditions combine. Around 9:00 CT, return variability and the spread are both up while the count of order-book events is down; the paper's label for that combination is temporary illiquidity. In the final five minutes before the 15:00 CT cash close, order-flow variability, average event size and depth are all up while return variability and the spread are down; the label there is high liquidity.
The two papers do not line up on that last point. The volatility-curve study has volatility still rising into the 15:00 CT equity close, while the order-book study has return variability declining in the final five minutes before it. Their samples, data frequencies and measures all differ — fifteen years of 30-second transaction prices against six years of quotes stamped to the second — and neither result is a correction of the other.
What happens to liquidity around a scheduled economic release?
Price impact is already rising an interval before a scheduled release reaches the screen, and order flow quietens in the minutes ahead of it. Takahashi compares its estimates on days carrying one of eight US macroeconomic releases — the ISM manufacturing index, existing home sales and consumer confidence among them, borrowed from an earlier study of high-frequency trading — against days with nothing on the calendar, then puts release dummies into a regression. Most of the eight were published at 9:00 Chicago time, 10:00 Eastern, during the sample, with a handful of dated exceptions. None of the eight is the payrolls report or CPI, so none of what follows is evidence about those two.
Before the number: in the 8:55–9:00 interval the variability of order-flow innovations falls, which the paper reads as orders being withheld ahead of the release. The regression puts the rise in price impact one interval early too, and the drop in order flow's responsiveness to price both one interval early and in the release interval itself.
At the number: price impact is higher around 9:00 and flow responsiveness lower, and the variability of return innovations spikes in the 9:00–9:05 window. Releases that came in under forecasters' expectations pushed price impact higher still and flow responsiveness lower still. Adding the announcement dummies barely changes the model's adjusted R², which the paper takes as a sign that its market-activity variables already carry much of what a release does.
For release times on the Eastern calendar, and for how often prices jump on announcement days, our post on trading futures during news events covers the evidence. There is no news-trading restriction here, per the trading hours article, so whether to be positioned into a release is the trader's call.
What does session timing mean on a funded account?
On a simulated funded account with a daily flat time, the clock is a deadline first. Everything has to be closed by 4:45 PM ET, 45 minutes after the 3:00 PM CT equity close that the volatility curve is still climbing into. A position carried into that stretch has to be unwound inside the session, and anything open at the deadline is closed for you, with the result landing on the balance and the drawdown.
The hours after the 6:00 PM ET reopen are available, and they belong to the next trading day for counting winning days. They are also the lowest stretch of the curve: the volatility study puts the Asian block's diurnal effect well below the American one, even though the share of total volatility falling in Asian hours rose across the sample.
That gap in scale is where session timing becomes a sizing question. The same stop measured in points is not the same quantity of volatility at 9:00 CT as at 1:00 AM CT, which is the problem average true range exists to measure, and how many contracts fit behind that stop is arithmetic on the account's loss limit — our post on futures position sizing works it through.
What neither study settles is which stretch to pick. The order-book paper never leaves US cash hours, so it offers no overnight comparison, and neither measured what a trade placed in either stretch returned.
FAQ
What time does the futures trading day start and end?
On an account here, the session reopens at 6:00 PM ET and all positions have to be closed by 4:45 PM ET the following afternoon, with no trading between those two times; after a Friday the reopen falls on Sunday evening. On a shortened holiday session the flat time is 12:45 PM ET instead. The exchange's own calendar is separate from that rule and changes, so check it rather than a blog post.
Is the overnight futures session quieter than the US session?
In the E-mini S&P 500, yes, on the one measure read for this post. Andersen, Su, Todorov and Zhang estimate the day's volatility curve over 2005–2020 and find the diurnal effect far smaller in Asian business hours than in American ones, by a multiple they do not quantify, even though the same U-shape turns up inside each block. Their subsample comparison also has Asian hours taking a larger share of total volatility by the end of those fifteen years.
Why is the last five minutes before the cash close different?
In Takahashi's order-book study, the 14:55–15:00 CT interval combines readings that pull in opposite directions: depth is up and the average spread is down, while order-flow variability and the average size of an order-book event are up and the variability of returns is down. The paper's label for that combination is high liquidity. Its estimate of how strongly order flow answers a price move also surges over the final 14:45–15:00 interval, having bottomed out around 12:45–13:00.
Is it safer to trade when the market is quiet?
Neither study read for this post measured returns or risk-adjusted outcomes by time of day, so there is no result here to lean on. What the order-book study does report is that the same imbalance in order flow pushes price further as depth falls away, and that around 9:00 CT — when most of the releases it examines came out — the spread widens while order-book event counts drop. A quiet tape is not automatically a forgiving one.
Sources
- Torben G. Andersen, Tao Su, Viktor Todorov and Zhiyuan Zhang — Intraday Periodic Volatility Curves (working version dated 24 January 2023)
- IDEAS/RePEc — journal record for Intraday Periodic Volatility Curves, Journal of the American Statistical Association 119(546), 2024, 1181–1191
- Makoto Takahashi — Returns and Order Flow Imbalances: Intraday Dynamics and Macroeconomic News Effects (arXiv:2508.06788v4, 8 October 2025)
- Lumen Futures Help Center — Trading hours
- Lumen Futures Help Center — Winning days
Educational content about futures markets and simulated trading. Not investment advice, and not a solicitation to trade. Trading futures involves substantial risk of loss. Read the full risk disclosure.