CME energy futures stop printing at 17:00 New York time on Friday and resume at 18:00 on Sunday. For 49 hours a week, oil has no official price.
This spring that gap became part of President Trump's broader negotiation tactics. US air strikes against Iran began on Saturday 28 February 2026, and for six consecutive weekends the war's biggest developments landed while the oil market, and most markets with it, stood closed. Through all of them, a WTI perpetual on Hyperliquid's xyz markets kept trading. Every fill on the venue is public and labelled, so those weekends produced something new: a trade-by-trade record of price formation with the professional market shut.
We pulled the record, roughly 39 million fills on the WTI contract, and went through it.
- The weekend the war began, the live oil price was set by 605 traders doing $14 million of volume. A week later the same market did $306 million; by mid-March, over half a billion.
- On the blow-off weekend, oil drifted +6.1% while CME was closed, spiked to $118.00 within hours of the reopen, and fell 32% from that print by Monday evening.
- Across 21 weekends, the weekend's direction matched the next session's direction 41% of the time. On average, 88 cents of every dollar of weekend drift was given back within 24 hours of the reopen.
- The same trade size moved the weekend market 2.4 times further than the weekday market.
- Average weekend funding fell from 71% annualised on the war's first weekend to 6% on the sixth, while the war itself escalated.
- On the ultimatum weekend, 75% of everyone trading oil was an account created that weekend. 94.5% of those accounts have lost money to date.
A market appears in seven days
Before the war, the WTI perpetual's weekends averaged $1.9 million of volume across 159 traders. That was the market on duty when the strikes began on the night of 28 February, with tankers stalling behind the Strait of Hormuz: 605 people, $14 million, and a price that climbed 4.5% by Sunday night.
It took about a week for the market to catch up with the war. Word spread through feeds and group chats, and the second war weekend did $306 million, more than twenty times the first. By mid-March, with the new supreme leader declaring Hormuz shut, a single weekend traded $536 million. It is safe to assume that the demand for a weekend oil price had existed all along; what had been missing was a place to express it, and once traders found one, it filled in days.
| Weekend | Backdrop | Weekend drift | Next 24h after reopen | Volume | Traders (first-time) |
|---|---|---|---|---|---|
| 28 Feb – 1 Mar | War begins | +4.5% | +0.8% | $14M | 605 (54) |
| 7 – 8 Mar | Strikes intensify | +6.1% | −11.1% | $306M | 5,791 (1,280) |
| 14 – 15 Mar | Hormuz closure | +4.5% | −7.8% | $536M | 7,385 (1,039) |
| 21 – 22 Mar | 48-hour ultimatum | +2.9% | −11.9% | $226M | 14,377 (10,836) |
| 28 – 29 Mar | Houthis join | +1.8% | +2.0% | $257M | 4,069 (397) |
| 4 – 5 Apr | Ceasefire talks | +0.6% | −0.1% | $228M | 9,364 (5,922) |
Each of those weekends had a specific trigger, and each landed while the market was shut:
- 28 Feb – 1 Mar: CENTCOM strikes began on the Saturday. Iran's supreme leader was killed on the first day of operations, and some 150 freight ships, many of them tankers, stalled behind the Strait of Hormuz.
- 7 – 8 Mar: Trump posted that "today Iran will be hit very hard" on the Saturday morning. Blasts hit Bahrain, Qatar and the UAE hours after Iran's president had apologised to its Gulf neighbours, and reports circulated that Iran would name a new supreme leader within 24 hours.
- 14 – 15 Mar: Iran's new supreme leader declared the Strait of Hormuz should remain closed, and the IRGC declared any US, Israeli or allied vessel transiting it a "legitimate target".
- 21 – 22 Mar: a 48-hour ultimatum from Trump to reopen Hormuz, backed by a threat to "obliterate" Iranian power plants, expired over the weekend. Iranian missiles struck near Israel's main nuclear research centre on the Saturday night.
- 28 – 29 Mar: the Houthis entered the war with ballistic missile launches at Israel, opening a second maritime chokepoint risk at Bab el-Mandeb.
- 4 – 5 Apr: Iran rejected a 48-hour ceasefire proposal while a 45-day framework circulated through mediators. The ceasefire came the following Wednesday.
The infrastructure held throughout. Across swings of twenty percent, hourly funding prints as extreme as +733% and −792% annualised, and weekends with tens of millions in forced liquidations, the tape is continuous. Settlement, margining and liquidation processing ran through a war without an interruption we can find in the data.
The blow-off
WTI closed the CME week at $90.67. Then came the Saturday-morning "hit very hard" post and the supreme-leader succession reports, and over the closed weekend the perpetual ground up 6.1% to about $95.80. When CME reopened on Sunday evening the move accelerated instead of correcting: the perpetual printed $118.00 in the early hours of Monday UTC, 23% above where the weekend left it, then fell to $80.15 by Monday evening. It finished the first full session 11% below the reopen price. More than $30 million of short positions were liquidated on the way up. The traders who were right about the destination were carried out before the market got there.
Interestingly, the top printed hours after the real market reopened, not during the weekend, and that ordering holds across the whole sample. In the first hour after the reopen the market tended to continue the weekend's direction; the reversal completed only over the following session. Regressing the post-reopen move on the weekend drift gives a slope of +0.28 at one hour, +0.35 at four hours, and −0.88 at 24 hours. Sunday-evening futures are thin and trade the same nervous money that traded the weekend. The deep session, London and New York desks, hedgers, calibrated views of physical barrels, arrives Monday, and that is when the weekend's swings seem to get repriced.
Two measurements
During the war, a million dollars of trading moved the weekend market 16 basis points, against 6.7 on weekdays. The ratio held at 2.4 through the post-war weeks as well, so it is a property of the weekend rather than of the war. The explanation is ordinary enough: the professional liquidity that absorbs flow during the week is largely absent on a Saturday, and the same order moves the price further on a thinner book.
Direction carried little information either. The weekend's move agreed with the following session's only 41% of the time, and the correlation between weekend drift and the Monday move was negative, at −0.51. Larger weekend moves were somewhat more likely to reverse than to continue.
On Monday 23 March, after a weekend under an expiring 48-hour ultimatum, reports of productive talks broke and oil fell more than 9%. Iran's parliament speaker dismissed the reports as fake news designed to manipulate the financial and oil markets. The tape shows the same sequence as the other war weekends: a drift upward while the market was closed, repriced once the full session arrived.
The market learned in public
Funding, the hourly payment between longs and shorts that tethers a perpetual to its reference price, works as a meter of what the crowd pays to hold a position.
On the war's first weekend, longs paid an average of 71% annualised to hold oil with no anchor. The second weekend cost 57%, then 42%, 20%, 11%, and finally 6%. The decline is monotonic, and it happened while the war escalated. The panic premium decayed in six weeks because the headlines stopped being new, and because every basis point of weekend overshoot was a payment available to anyone calibrated enough to collect it.
The crowd also priced some things well. The spread between the venue's Brent and WTI contracts widened from about $1 in early March to $9.43 going into the ultimatum weekend, which is the fingerprint a Gulf supply shock should leave: seaborne barrels repricing against landlocked ones. The magnitude was wrong; the structure was right.
The cost of the learning fell on the newest accounts. Weekend-only WTI traders during the war had a median trade size of $11 and a median account value of $93, ran 13x average leverage, and 96% of them had never traded the venue before the war began. Accounts that traded through the week, with median tickets around $400, collectively earned about $5.4 million on the war weekends. The weekend-only cohort lost money as a group, and 94% of its members are down overall.
Where this goes
The weekend oil price of early 2026 existed, cleared, and settled through a war. It was also thin, over-amplified, with most of its moves handed back within a session.
But a price with knowable biases is still a price. Before this market existed, a treasurer or a physical trader watching tankers stall behind Hormuz on a Saturday had no way to act until Sunday evening. Now there is a venue where that risk trades at any hour.
Nxos builds the regulated rails that connect operators who run their books in fiat to this class of market. Reading these venues at the fill level, including how young they are and in which direction they trend, is part of that work.