- 2 alignedVenues
- L2 bookPerp data
- Per-timestampAlignment
- BTC /
ETH / SOL / XRPSymbols
A Polymarket BTC Up/Down contract and a Hyperliquid BTC perpetual price the same underlying with completely different instruments. When BTC moves, which one reprices first? That is a price-discovery question, and answering it needs both books on the same clock. We keep both.
The cross-venue study is one of the few you genuinely cannot do with public data alone, because it needs synchronized history from two different venues. That is exactly the kind of question worth keeping the data for.
What you are joining
Polymarket odds
The prediction-market snapshots, implied probability, spread, and full depth for each Up/Down contract.
- 0-1 probability
- Millisecond stamps
- BTC/ETH/SOL/XRP
Hyperliquid perps
The perpetual futures L2 book for the same underlying, best bid/ask, mid, and depth, once per second.
- L2 order book
- Same symbols
- Scale tier and up
Both feeds stamp a millisecond timestamp from the same capture pipeline, so they line up on a shared time axis with no guesswork. The prediction-market side carries the implied odds; the perp side carries the underlying price, two views of the same BTC move, ready to overlay.
Two instruments, one underlying
The reason this comparison is interesting is that the two venues express the same BTC move in completely different units. A Hyperliquid perp quotes a price in dollars that moves continuously; a Polymarket Up/Down contract quotes a probability between 0 and 1 that has to translate that same move into the odds of finishing above a strike. They are not redundant feeds, they are two different lenses on one event, and the gap between them is where the signal lives.
- PerpContinuous price, deep book, ~1/sec
- ContractProbability 0-1, thinner book, ~20 Hz
- StrikeThe level that ties the two together
- One clockMillisecond stamps from one pipeline
Putting them on one clock
Align on the nearest prior timestamp
The correct way to merge two feeds is to attach, to each prediction-market snapshot, the most recent perp price that preceded it, never one from the future. Done that way, the join is clean and the lead-lag analysis that follows is trustworthy. Done carelessly, you leak tomorrow’s price into today and every result is wrong.
Which venue moves first?
With the venues aligned, the question becomes measurable: does a move in the perp tend to precede the change in Polymarket odds, or the other way around? Scan across a range of small time offsets and find where the relationship is strongest. If the perp leads, a perp move is an early read on where the odds are heading. If the prediction market leads, the contract is the price-discovery venue for that event.
Lead-lag read
Whichever venue consistently moves first becomes an early signal for the other.
- Perp leads → early odds read
- Odds lead → contract is price-setter
- Measured, not assumed
Dislocations
Persistent gaps between the perp-implied direction and the contract odds are candidate arbitrage.
- Net of spread crossed
- Watch the thinner book
- Confirm with depth
Sizing reality
The perp book is deep; the short-dated contract often is not. Size against the prediction-market side.
- Cap to contract depth
- Model real slippage
- Avoid moving the book
The data is portable by design
Both feeds come out as plain, time-stamped records, nothing proprietary, nothing locked in. Pull them into whatever you already use, line them up on the shared clock, and the analysis carries over unchanged. The whole pitch is that the data fits your workflow, not the other way around.
Where a cross-venue study goes wrong
The join is simple, which is exactly why it is easy to get subtly wrong. A handful of mistakes turn a real lead-lag finding into an artifact, and every one of them is avoidable if you know to look for it.
- 1Forward leakage, attaching a perp price that came after the snapshot. Always merge on the nearest prior timestamp, never the nearest in either direction.
- 2Mismatched clocks, comparing two feeds stamped by different systems. Both feeds here share one capture pipeline, so the clock is the same; trust it.
- 3Sampling mismatch, the perp arrives ~1/sec, the contract at ~20 Hz. Resample deliberately rather than letting one feed’s density distort the other.
- 4Ignoring the spread, a dislocation that does not survive crossing the spread on the thinner book is not an opportunity, it is a quote artifact.
- 5Sizing to the wrong book, the perp is deep, the contract usually is not. Cap your size to the prediction-market depth or you are trading a number you cannot fill.
Two venues, one clock, no leakage, get those three right and the lead-lag answer is trustworthy; get any one wrong and the whole study is fiction.
Get both feeds aligned
See the Hyperliquid perpetual data alongside the Polymarket snapshots, and check which tier unlocks exchange access.
Frequently asked questions
What Hyperliquid data does Resolved Markets capture?
Resolved Markets captures the Hyperliquid perpetual futures L2 order book for BTC, ETH, SOL, and XRP, via an event-driven WebSocket feed sampled and stored once per second. Each record carries best bid/ask, mid, spread, and depth totals with a millisecond timestamp, so it aligns cleanly with the Polymarket snapshots.
Which tier do I need for exchange data?
Hyperliquid perpetual data is available on the Scale tier ($49/month) and Enterprise. The Free and Pro tiers focus on Polymarket prediction-market data. Scale also unlocks all other market categories, sports, weather, economics, social, and equities.
Can I really do cross-venue lead-lag analysis with this?
Yes, that is exactly what time-aligned storage is for. Because both the Polymarket snapshots and the Hyperliquid book carry millisecond timestamps from the same capture pipeline, they merge onto one clock with no forward leakage, and a lead-lag scan tells you which venue tends to move first.
How should I size a cross-venue trade?
To the thinner book, which is almost always the prediction-market side. The Hyperliquid perp book is deep, but a short-dated Up/Down contract often is not, so the contract depth is your real constraint. Cap position size to the resting depth visible in the snapshot, model the slippage you would take crossing the spread, and never assume you can fill the size the perp book suggests on the contract side.