Oil futures close for 49 hours every weekend, from Friday evening to Sunday evening New York time. Perpetual contracts on Hyperliquid never close. Across 30 weekend closures this year, the move in the WTI perpetual explains about three quarters of where futures reopened. Nothing was trading for it to copy, so it was establishing the price on its own.
The war began inside one of those windows, on Saturday 28 February, and has produced weekend news ever since. WTI and Brent have traded $74bn and $36bn since March, with open interest peaking above $600m and $550m. A perpetual has no expiry, so a hedger holds one position instead of a series of expiring ones, and a funding rate paid between longs and shorts keeps it near the price it tracks.
The CL perpetual misses the futures reopen by about a percentage point, so the reading is only usable when oil moves more than that. Oil cleared a point across twenty of the thirty closures this year, and the perpetual got the direction right in all twenty. In the other ten it was wrong three times. Volume seems to be what separates them: eighteen of the twenty large-move closures saw over $100m trade on the perpetual, against one of the other ten.
At its best the match is very close. Over the closure of 8 May the perpetual rose 2.97% and futures reopened 2.90% higher, a difference of less than a tenth of a point.
The largest miss was the war's first weekend. Markets were shut over 27 February, during which Ayatollah Ali Khamenei was killed. The perpetual rose 4.6% to finish around $70, but futures reopened at $75 and only came back to $71 by the close. The perpetual never followed the opening spike. Across all thirty closures it tracks the opening print far more closely than the settle, which is why the open is the benchmark used here.
August followed that pattern. Weekend volume collapsed from an April peak of $956m to $63m and then $37m across the closures of 14 and 21 August, and on both the perpetual moved less than a tenth of a percent. Its direction matched the reopen each time, but a move that small sits inside its own margin of error. On 28 August volume recovered to $96m, the perpetual moved 2.0%, and futures reopened 1.5% higher.
Orders below $100k fill without moving the price, whether futures are open or shut. Above $1m, spreading the order over the two hours of a weekend closure costs more than taking it in one go, because the book is thin enough that the price drifts further in that time than crossing it at once would have cost.
The cost here is slippage, how far the price moves against you while the order fills. On orders above $100k during a closure it has risen from 0.23 basis points in April to 3.19 in August, or about $30 to $320 on a $1m order. Midweek it has gone from nothing to 0.71. Filling the same order now takes 43 price levels instead of 30, because less is resting at each one, and orders of that size have dropped from 1,953 a month to 478.
April and May were the cheapest months to trade despite April being the most volatile, because the war brought enough participation to offset the volatility. As that participation left, the book thinned with it.
On CL orders worked over time the median drift is close to zero at small size, but the tenth to ninetieth percentile spans roughly 50 basis points in every size bucket.
Kalshi lists a ladder of settlement levels on oil, currently from $51 to $96 in small steps. Prices across that ladder describe the whole range of outcomes the market is paying for, the way an options chain does, which is something a perpetual cannot show at any single moment.
Polymarket is the bigger venue on oil price, but its most popular markets are the geopolitics driving those prices. Since January it has traded $317m on the oil price against Kalshi's $236m, $413m on the Strait of Hormuz against $57m, and $4.2bn on the wider conflict against $91m. Oil price contracts are therefore 61% of Kalshi's book but only 6% of Polymarket's.
Both venues spiked on the war and both retraced, but differently. Polymarket peaked at $545m in the ceasefire week of 6 April and now trades at 5% of that. Kalshi sits near its own peak, because its volume moved into daily and hourly price ladders that keep trading without an event: its oil ladders set a fourth straight weekly record at $32.6m, and in the week to 30 August Kalshi was the larger venue for the second week running, $34.1m against $29.1m.
On Polymarket, the contract on Hormuz traffic returning to normal by 31 December moves inversely to the oil price, correlating at about -0.4 day to day since May. Across the fourteen weekend closures in that window the odds moved every time, and their direction matched the reopen in twelve. But they explain 35% of the reopen against the perpetual's 95%, so as a price signal they add nothing to it.
Of the three instruments here, the CL perpetual on Hyperliquid is the most useful. It gives a crude price through the 49 hours when futures are shut, and above a one-point move it has called the direction every time. In practice that has meant weekend volume above $100m.
Thirty weekends is a short history, and the result is specific to that contract. The Brent perpetual moves about three times as far as Brent futures do over the same weekend, despite almost identical funding, which makes it unusable as a guide.
A US desk cannot trade it yet. Hyperliquid blocks US users. On 26 August the Hyperliquid Policy Center and trade[XYZ] asked the CFTC to allow energy perpetuals onshore, and on 31 August Bloomberg reported talks to reach US traders through Payward and its Bitnomial exchange, though that route covers crypto perpetuals and not these contracts. Neither has been approved.
As oil price proxies the prediction markets add little, and Kalshi, the one venue a US institution can reach directly, is the thinner of the two. Polymarket's Hormuz and conflict markets price a different variable, and nowhere else prices those outcomes at all, but its volume sits on the global platform US investors also cannot access.
Volume, open interest, funding, liquidations, trade sizes and prediction market volume are queried on Dune. Futures prices and the Bloomberg report are external. For informational purposes only; not investment advice or a recommendation on any instrument.
Looking to use Dune for your company?
