Kumo Research · Gold & oil
Gold and oil on weekends: does 24/7 on-chain trading predict the Sunday-evening futures reopen?
Gold and oil futures stop trading on Friday evening and reopen on Sunday at 6 pm New York time. On-chain markets – Hyperliquid's gold and oil perps and the PAXG gold token – keep trading all weekend. We tested whether those weekend prices tell you where the futures will reopen.
By Alex Wu · Published · 5 min read
Key findings
- For oil, the answer is a clear yes. By 5 pm on Sunday, the Hyperliquid WTI perp had the direction of the futures reopening gap right on 23 of 24 weekends (96%), with a correlation of 0.94 and an average error 64% smaller than assuming "no change".
- Oil's weekend news arrives on Saturday and Sunday, not Friday night: the perp was nearly flat at Saturday noon before the biggest gaps, and most of the move appeared by Sunday noon.
- Oil weekend gaps were large in 2026 – an average of 2.3% since April, and up to 8.6% (April 2026). In those weeks the on-chain price showed most of the gap hours before the futures reopened.
- For gold the signal is much weaker. Gold's typical reopening gap is small (0.3–0.5%), and weekend on-chain moves got the direction right only 62–71% of the time.
- Over two years of PAXG data, using the token's weekend move as a forecast made the error larger, not smaller: its weekend moves were mostly noise, bigger than the gaps they tried to predict.
Two markets that close, and two that don't
The main global benchmarks for gold and crude oil are futures traded on CME: COMEX gold (GC) and NYMEX WTI crude (CL). They trade almost around the clock from Sunday evening to Friday afternoon, but close from about 5 pm New York time on Friday to 6 pm on Sunday. When they reopen, the price often "gaps" to a new level to reflect anything that happened over the weekend.
On-chain markets never close. Hyperliquid lists perpetual futures on gold, silver, WTI and Brent crude through trade.xyz, and PAXG – a token backed by physical gold – trades 24/7 on crypto exchanges. If weekend traders on these markets react to real news, their prices should anticipate the Sunday reopening gap.
We tested this for:
- WTI oil: the Hyperliquid perp (xyz:CL), 24 weekends from April to October 2026.
- Gold: the Hyperliquid perp (xyz:GOLD) over the same 25 weekends, and the PAXG token on Binance over 113 weekends since June 2024.
For each weekend we measured the futures gap – Sunday's reopening price against Friday's last price – and compared it with the on-chain market's move from Friday 5 pm to Saturday noon, Sunday noon and Sunday 5 pm, one hour before the reopen.
Oil: the weekend price is a strong preview
Did the weekend on-chain price get the Sunday reopen direction right?
Share of weekends where the on-chain move pointed the same way as the futures reopening gap, % (New York time)
- Oil – Hyperliquid WTI perp
- Gold – Hyperliquid perp
- Gold – PAXG token (since 2024)
Source: Hyperliquid trade.xyz and Binance PAXG hourly prices vs CME gold and WTI futures (Yahoo Finance) · Chart: Kumo Research, kumocast.io
Oil stands out. At Saturday noon the Hyperliquid oil perp had no useful signal – it got the direction right only half the time. But by Sunday noon it was right on 79% of weekends, and by Sunday 5 pm on 96% (23 of 24). Its weekend move had a correlation of 0.94 with the actual reopening gap.
…and did it improve on "no change"?
Reduction in the average error vs assuming the futures reopen where they closed, % (negative = worse)
- Oil – Hyperliquid
- Gold – Hyperliquid
- Gold – PAXG
Source: Kumo Research calculation · Chart: Kumo Research, kumocast.io
As a forecast, the Sunday 5 pm oil perp price cut the average error by 64% compared with assuming the futures would reopen unchanged. Even at Sunday noon it cut the error almost in half.
Oil's news arrives at the weekend
Oil weekend gaps were unusually large in 2026: an average of 2.3% in absolute terms since April, compared with about 0.3–0.5% for gold. The table shows the biggest ones.
| Weekend (Friday) | Hyperliquid oil, Sat noon | Sun noon | Sun 5 pm | Futures reopening gap (Sun 6 pm) | Futures at 10 am Monday |
|---|---|---|---|---|---|
| 2026-04-10 | -0.1% | +5.5% | +7.6% | +8.6% | +5.3% |
| 2026-07-31 | +0.0% | -6.1% | -6.6% | -5.5% | -6.5% |
| 2026-07-24 | -3.5% | -5.9% | -4.7% | -5.2% | -5.7% |
| 2026-06-12 | -2.1% | -1.4% | -2.0% | -4.6% | -5.5% |
| 2026-06-05 | +1.8% | +2.7% | +3.0% | +3.1% | +0.6% |
| 2026-09-11 | -0.9% | +0.4% | +0.9% | +3.1% | +4.3% |
| 2026-05-29 | +0.4% | +1.1% | +1.1% | +2.9% | +7.8% |
| 2026-05-08 | -0.0% | -0.7% | +3.0% | +2.7% | +1.3% |
All changes from the futures' last Friday price (hour ending 5 pm New York time) and, for the perp, from its own price at 5 pm Friday.
In almost every case the perp was close to flat at Saturday noon and moved during Saturday afternoon or Sunday – the kind of timing you would expect if news arrives over the weekend itself. Producer-group decisions (OPEC+ has often announced output decisions at weekend meetings) and geopolitical events do not wait for markets to open. On the weekend of 10 April 2026, the perp was up 5.5% by Sunday noon and 7.6% by 5 pm; the futures reopened 8.6% higher. On 31 July, it was down 6.1% by Sunday noon; the futures reopened 5.5% lower.
Gold: a weak signal at best
Gold is different. Its weekend gaps are much smaller – an average of 0.31% over two years and 0.51% in the 2026 sample – because gold reacts less to weekend news than oil does.
- The Hyperliquid gold perp pointed the right way on 71% of weekends by Sunday 5 pm (correlation 0.60) and reduced the error by a modest 14%.
- PAXG, over a longer two-year sample, got the direction right on only 53–62% of weekends depending on the hour. Worse, using its weekend move as a forecast increased the error by 25–47%. PAXG's weekend moves (about 0.4% on average by Sunday) were larger than the actual gaps (0.31%), so most of what it did over the weekend was noise.
On the same 25 weekends in 2026, PAXG and the Hyperliquid gold perp performed almost identically. The difference between the two-year PAXG result and the 2026 result is the period, not the venue: 2026 weekends simply carried more news for gold.
| Market | Weekends | Typical reopening gap | Correlation at Sun 5 pm | Direction right at Sun 5 pm | Error reduction at Sun 5 pm |
|---|---|---|---|---|---|
| WTI oil – Hyperliquid (Apr–Oct 2026) | 24 | 2.27% | 0.94 | 96% | +64% |
| Gold – Hyperliquid (Apr–Oct 2026) | 25 | 0.51% | 0.60 | 71% | +14% |
| Gold – PAXG (Jun 2024–Oct 2026) | 113 | 0.31% | 0.44 | 62% | -47% |
Direction counts only gaps larger than 0.15%. Error reduction compares the on-chain implied gap with a "no change" guess.
What this means
- For oil, weekend on-chain prices have been a strong guide to where the futures reopen on Sunday evening – especially after big weekend news.
- Most of the information for oil arrives on Saturday afternoon and Sunday, so a Saturday-morning price says little.
- For gold, weekend on-chain moves are a weak and often misleading signal, because weekend gaps are small and on-chain weekend trading adds noise.
- In both cases, the on-chain markets add information mainly when there is real weekend news to price.
The oil and Hyperliquid gold samples cover only about six months of weekends, so these results should be read as early evidence. For live levels and volatility ranges, see our WTI oil, Brent oil and gold pages.
Method, data and limits
Data: hourly prices of COMEX gold futures (GC=F) and NYMEX WTI crude futures (CL=F) from Yahoo Finance; hourly closes of Hyperliquid trade.xyz perps xyz:GOLD and xyz:CL (from April 2026) and Binance PAXG/USDT (from November 2023). Futures sample from June 2024 to 30 October 2026.
Measures: futures gap = first Sunday price (bar starting 6 pm New York time) ÷ last Friday price (bar ending 5 pm) − 1. On-chain move = price at Saturday noon, Sunday noon or Sunday 5 pm ÷ price at Friday 5 pm − 1. Direction counts gaps above 0.15%. Error reduction = 1 − mean |gap − on-chain move| ÷ mean |gap|.
Limits: small samples for the Hyperliquid markets (24–25 weekends); futures roll dates can create artificial gaps; PAXG tracks spot gold, not futures; holiday weekends included where data exists. Not investment advice.
Download the data (CSV, CC BY 4.0)
Sources
How to cite
Kumo Research (2026). "Gold and oil on weekends: does 24/7 on-chain trading predict the Sunday-evening futures reopen?". Kumocast. https://www.kumocast.io/research/gold-oil-weekend-onchain-prices
This study describes past market data. It is not investment advice and past behaviour does not guarantee future results.
Live data for these markets
More from Kumo Research
- Gold & oilGold vs Bitcoin in the 2025–2026 sell-offs: which one actually held up?5 min read →
- BitcoinIs Bitcoin getting less volatile? 15 years of data, and how BTC now compares with Tesla, Nvidia and gold5 min read →
- Stock perpsWhat does it cost to hold a stock perp? A year of funding rates on Tesla, Nvidia, Apple and other on-chain stocks6 min read →
About the researcher
Alex Wu
Crypto, DeFi, perps and FX · driven by data
Alex designed the Kumocast model behind these signals, sets its rules, reviews its accuracy record and updates the methodology whenever the model changes. Signals are refreshed every 30 minutes with live market data.