Data
Hyperliquid, on the record. liquary.xyz/data is a public reference surface: datasets we keep because the chain does not. Hyperliquid purges a HIP-4 market shortly after it settles, and serves validator votes only while an action is in flight; Liquary keeps the record.
There is no wallet and nothing to connect: every page is readable, linkable and citable by anyone.
hip-4.org redirects to the HIP-4 dataset
(liquary.xyz/data/hip-4). Every route on it lands on the dataset’s front
page: to link a specific market, deployer or validator, use its
liquary.xyz/data/... URL. It is not an API host.
What’s in it
The HIP-4 dataset
Every prediction market since the first one: traded volume, settlement outcomes, validator governance, and the written reason behind each resolution.
- Volume over time, flat or split by category, daily or cumulative.
- All markets, searchable and filterable, each linking to its own record.
- Category share, biggest markets, and the validators card.
A market’s record
One page per settled market: its odds curve from first trade to settlement, the validators who signed with their stake weight and delay, the deployer’s written settlement reason, the rules, the final score where there was one, and the rest of its family.
A tournament question has many legs: the family’s volume is the sum of them all, a single leg’s is much smaller. A market record prints both, labelled.
Deployers
Who creates each market and signs its settlement. Per deployer: markets listed, how many are active, how many settled, how fast they settle (median, with a distribution), and volume. Third-party deployers appear automatically as they launch their first market.
Volume is what people traded plus what the protocol paid out at settlement; a toggle removes the payout. Mints, merges and negations are in neither figure, and are charted separately.
Validators
A card on the dataset’s overview lists every validator that signs HIP-4 settlements, with its stake, settlements signed, participation rate and median delay.
These figures are comparable to each other, not to anything outside them. Rosters are measured over Liquary’s capture window; settlements that finalised before capture started are counted as gaps, and the pages say how many.
Governance
liquary.xyz/data/governance is the second dataset: every L1 governance action Hyperliquid’s validators vote on (new market templates, listings, settlements), with who voted and the stake each carried at that moment. Three views: what is in flight now, the history, and who votes, validator by validator. The page warns when the capture has not reached Hyperliquid recently.
Research
Questions answered over the whole settled HIP-4 archive:
- Market calibration: what the market implied, graded against what actually happened.
- Where the money is: how concentrated the volume is across markets.
- Who settles fast: how long each deployer takes to settle after expiry.
- The biggest upsets: the markets where the cheap side won.
Reading it honestly
- Markets are marked complete or partial depending on how much of their life was captured.
- A market that settled before capture began says its rules are missing rather than inventing them.
- On some settled markets a share of the trades happened after settlement (the winner converging to $1, the loser draining), and they are labelled as such.
The same data, as JSON
For every settled market, the record behind its page (the result, the settlement and its voters, the volumes and the traders) is served by a free, unauthenticated, CORS-open API, back to May 2026. The odds curve is drawn here only. See the Archive API.
Open the HIP-4 dataset in Liquaryliquary.xyz/data/hip-4