On Feb. 28, coordinated strikes hit Iranian nuclear facilities while most benchmark commodity markets sat dark.
Traditional gold futures on CME’s COMEX exchange wouldn’t reopen until Sunday evening Central Time, leaving a 48-hour window where macro risk had nowhere obvious to express itself.
Except it did: on venues that never close.
By the time COMEX gold futures flickered back online Sunday at 5:00 PM CT, perpetual futures contracts tracking gold and silver on always-on derivatives platforms had already written the first draft of Monday’s gap.
Traders didn’t wait for permission. They repriced geopolitical risk in real time, using whichever venue accepted their orders, and when the benchmark finally opened, it caught up to a price that had been forming all weekend.
This isn’t a story about decentralized finance replacing traditional exchanges. It’s about continuity.
Markets exist to discover prices in the face of uncertainty. When benchmark futures close, the best tradable proxy becomes the weekend risk barometer. Always-on derivatives don’t need larger open interest than COMEX to matter. They need to be open, tradable, and informative under stress.
The advantage isn’t purity, but uptime.
Testing the weekend tape
What happened during that closure window offers a case study in how price discovery relocates when reference markets go dark.
Under normal weekday conditions, perpetual contracts trade on a structural basis relative to front-month futures.
Front-month contracts embed the cost of carry, and perpetuals track the spot price more closely through funding, which is the periodic payment between long and short positions that pins the perpetual price to the underlying.
A modest, persistent gap between the two is expected.
However, the weekend of the Iran strikes created an experiment. With COMEX futures offline from Friday’s 4:00 PM CT close until Sunday’s 5:00 PM reopen, gold and silver perpetuals on platforms like Hyperliquid and Binance became the only liquid venue for expressing macro risk in precious metals.
Both platforms list 24/7 perpetual contracts tied to gold and silver, giving traders continuous access to metals exposure.
Analyst Kunal Doshi measured what happened during peak volatility hours.
Hyperliquid’s gold and silver perpetuals are priced at a median premium of roughly 75 to 78 basis points above Binance’s equivalent contracts.
More importantly, when COMEX reopened, Hyperliquid’s weekend price sat closer to the first benchmark print than Binance’s tape by approximately 22 to 31 basis points.
The weekend market that led turned out to be the one that better predicted the gap.
Those measurements don’t prove causation, but they reveal something about microstructure under stress. The CME’s reopening process includes an Indicative Opening Price period followed by a no-cancel lockdown phase immediately before trading resumes.
That makes the first tradable print after resolution a meaningful benchmark for whether the weekend tape accurately drafted where risk needed to land. In this case, it did.
Why continuous markets can lead
Multiple mechanisms explain why an always-on venue might generate useful price signals even when benchmark liquidity dwarfs it during normal hours.
Continuity beats size when the reference is closed. The open market becomes the marginal venue for first-response risk expression.
Traders holding positions over the weekend or needing to hedge breaking news can’t wait for Sunday evening. They route to whatever accepts orders.
Reopen microstructure creates a discrete event that continuous markets can anticipate.
CME Globex’s pre-open mechanics, such as IOP calculation, lockdown period, and opening resolution, turn the reopen into a moment.
Continuous venues sketch the path toward that moment in real time, producing a signal that legacy markets either validate or correct when they resume.
Positioning telemetry runs live. Funding rates reveal the price of leverage in real time. When funding flips sharply positive or negative, it signals where pressure lies and which side must pay for the privilege of staying in the trade.
Open interest shifts without waiting for Monday. That information feeds back into price before benchmarks reopen.
Global participation changes the weekend cohort. The weekend tape isn’t just absent from US institutional desks. It’s different time zones, different hedgers, different urgency profiles showing up when the primary venue is dark.
That mix might be less deep, but it’s not necessarily less informed about macro shocks hitting during off-hours.
Operational risk matters more than participants assume. Even “always-on” legacy infrastructure can go offline unexpectedly. CME metals futures experienced an outage on Feb. 25, reminding traders that benchmark status doesn’t guarantee access.
The platforms that actually stayed live during that window became the only venue for price discovery, whether they were designed for that role or not.
But don’t overread one weekend
A weekend can reveal a market’s reflexes, but it doesn’t settle the verdict.
Perpetuals aren’t futures. Index construction, mark price methodology, and funding mechanics can distort the price signal in ways that don’t show up in a simple premium comparison.
Liquidity can be optical: spreads look tight until depth disappears, and stress tests often reveal that bid-ask stability during calm periods doesn’t hold when everyone needs the same side.
Volume can lie. High volume per unit of open interest often reflects churn or recycling rather than new conviction. Doshi himself flagged this concern: if the same positions flip back and forth, the tape might look active without actually incorporating new information.
One weekend isn’t a law, and broader sampling complicates the narrative.
Blockworks analyzed Hyperliquid’s builder-deployed equity perpetuals and found that weekend “pre-open mid” prices came closer to the Monday reopen only about 50.7% of the time, with a median improvement of roughly 0.4 basis points.
That suggests the gold and silver performance during the Iranian weekend might be more about the specific asset class, the specific shock, and the specific participant mix than a generalizable advantage.
Markets that run continuously don’t automatically generate better signals. They generate different signals, and whether those signals prove useful depends on depth, participant sophistication, and how closely the contract design tracks the underlying benchmark.
| Measure | What it indicates (if clean) | How it can mislead | What to sanity-check |
|---|---|---|---|
| Perp–futures basis | Carry vs funding effects; how perp tracks spot vs front-month futures | Comparing unlike contracts (carry-embedded futures vs spot-anchored perp) can look like “signal” | Normalize vs spot; adjust for carry; compare basis during overlapping weekday hours as baseline |
| Funding rate | Directional pressure / “price of leverage” in real time | Can flip from mechanical imbalances (hedging flow, inventory) rather than new information | Compare funding shifts to price moves + open interest change; check persistence (hours, not minutes) |
| Open interest (OI) | Conviction / position build or unwind | OI can stay flat while participants churn; OI can be capped/managed by venue rules | Pair OI with liquidations + funding; look at changes by session (weekend vs weekday) |
| Volume | Activity / responsiveness to news | Can be recycling/churn (same risk flipping hands) rather than fresh conviction | Use volume ÷ OI; check trade size distribution; look for volume spikes without OI change |
| Spreads (top-of-book) | Instant liquidity / transaction cost | Optical: tight spreads with shallow depth; spreads stable until they suddenly gap | Add depth-at-1bp/5bp; average size-to-fill; slippage on market orders during the shock window |
| Mark price / oracle design | Stability; reduces manipulation; affects liquidations | Mark can lag real trades or smooth moves; different venues compute differently | Compare last vs mid vs mark; note oracle inputs and update cadence; check liquidation triggers vs mark |
| Reopen “first print” | Benchmark convergence; whether weekend tape “drafted the gap” | Pre-open mechanics can distort what counts as “first” (indications, resolution) | Define reopen anchor consistently (post-resolution tradable print); use the same candle alignment for all venues |
| One-weekend effect | Reflexes under stress; a stress test snapshot | Not generalizable; event-specific cohort/liq conditions dominate | Compare multiple weekends / shocks; separate “headline weekends” from normal weekends |
| Blockworks equity-perp sample | Base rate reality check on weekend predictive power | Different asset class/shock; builder-deployed equity perps may behave differently than metals | Note: ~50.7% closer to Monday reopen; ~0.4 bps median improvement; treat metals case as a special episode, not a rule |
Who owns Sunday night
The scale involved isn’t trivial anymore. Hyperliquid currently shows over $5 billion in perpetual open interest and processes billions in daily volume.
The platform’s HIP-3 mechanism allows builders to deploy new perpetual markets if they maintain 500,000 staked HYPE tokens and face validator-enforced slashing for malicious operation.
Open interest caps and other guardrails attempt to manage risk, but the core feature is permissionless market creation with continuous uptime.
Mainstream financial media noticed. MarketWatch explicitly reported traders using venues like Hyperliquid to gauge where crude oil might open after the weekend attack.
Bloomberg framed always-on perpetuals as the 24/7 hedge venue for oil, gold, and silver amid escalating tensions with Iran.
These aren’t crypto publications hyping native infrastructure, but traditional outlets acknowledging that price discovery has relocated because the benchmark was closed and risk needed to be expressed.
If always-on venues become the consistent first responder for weekend macro shocks, traditional exchanges increasingly become the settlement and reference reopen.
That changes who sets the narrative on Mondays. Instead of “markets gapped on news,” the frame becomes “markets caught up to the price already forming.” The gap was already drafted. The benchmark validated or corrected it.
CME itself understands the competitive dimension.
The exchange has moved toward 24/7 access in cryptocurrency derivatives, explicitly citing demand for always-on trading. Hours are now a product feature rather than an operational constraint.
The question now is which assets will develop reliable 24/7 shadow prices next, and whether those shadow prices prove informative enough for participants to trust them when the benchmark is dark.
The reopening is starting before it actually opens. That’s not ideology, but infrastructure responding to the reality that geopolitical risk doesn’t respect CME Globex maintenance windows.
The market that doesn’t sleep is becoming the market that explains the gap, and legacy venues will either extend their hours or accept that someone else writes the first draft of Monday.