1 Create
A YES/NO pair is minted on L1.
An operator calls registerTokensAndStandaloneOutcome on HyperCore. Two #-prefixed spot tokens are minted — one per side — opening near 50/50 unless seeded.
YES = #(id × 10) · NO = #(id × 10 + 1)
HIP-4 · Outcome Trading
HIP-4 outcome books. Testnet
Polymarket-style markets on the same L1 as Hyperliquid's perps. Fully collateralized, settled in a single block by the protocol — not a separate venue, not an oracle round-trip. Live on testnet today; no mainnet spec yet.
How HIP-4 Works
Three phases, all on L1, no off-chain matching.
HIP-4 markets share HyperCore's CLOB, matching engine, and account system with spot and perps. Pair
minting keeps YES + NO priced to exactly 1.00 — so the YES mark is the implied probability.
2 Trade
Either leg trades on the standard CLOB.
Pair minting mirrors the books — a buy on YES at 0.40 reflects an ask on NO at 0.60. Collateral locks until settlement; no leverage, no liquidation path, no funding to pay.
YES + NO = 1.00 USDH3 Settle
One block, no claim, no dispute.
At expiry an operator submits VoteGlobalAction with a settle fraction of 0 or 1. Every position resolves in a single L1 block — winner credited 1.00 per token, loser zero.
Protocol-driven settlement
The YES mark is the implied probability. Pair minting forces YES + NO to 1.00,
so the YES leg's price reads directly as the book's belief — no oracle, no inference. A YES at
0.65 means the market is pricing 65%.
Three shapes are live on testnet: Custom (multi-named questions like "who wins the race"), priceBinary (above/below a strike at expiry), and Recurring (auto-respawned priceBinaries every 15m, 1h, or 1d). All three trade as ordinary # spot pairs once minted.
Three shapes are live on testnet: Custom (multi-named questions like "who wins the race"), priceBinary (above/below a strike at expiry), and Recurring (auto-respawned priceBinaries every 15m, 1h, or 1d). All three trade as ordinary # spot pairs once minted.
Testnet Outcomes
Every outcome currently registered on testnet.
Pulled live from the outcomeMeta endpoint and joined to spotMetaAndAssetCtxs
for implied probability and 24h volume. Filter by ticker, question, or outcome shape.
| Outcome | Type | Implied Prob (YES) | 24h Volume | ID |
|---|---|---|---|---|
| Loading… | — | — | — | Loading outcomes from Hyperliquid testnet. |
How It Stacks Up
HIP-4 vs Polymarket vs Kalshi.
Three different bets on what a prediction-market venue should look like. HIP-4 is the
only one that puts matching, settlement, and collateral on the same L1 as a major perp book.
| Feature | HIP-4 (Hyperliquid) | Polymarket | Kalshi |
|---|---|---|---|
| Type | Native L1 CLOB | Hybrid (off-chain match, on-chain settle) | Centralized |
| Settlement | Auto via VoteGlobalAction, instant | UMA optimistic oracle, dispute window | Platform confirms |
| Chain | HyperCore L1 (~0.07s blocks) | Polygon (ERC1155 CTF) | Off-chain |
| Pair minting | Native in matching engine | Conditional Token Framework (smart contract) | Internal ledger |
| Oracle | On-chain price feed (system wallets) | UMA Optimistic Oracle | Internal |
| Dispute | None for price-linked; trust-based for subjective | UMA bond + vote | Platform decision |
| Composability | Same engine as perps + spot, unified margin | Isolated on Polygon | None |
| Cost | Near-zero gas | Polygon gas + venue fee | Platform fees |
| Regulation | Unregulated | Unregulated (non-US) | CFTC-regulated |
What This Means For Compost
The capital formation thesis extends to outcome books.
Compost is a capital formation layer for HIP-3 builder markets. HIP-4 is the same engine
producing a different shape, and the most interesting parts aren't the LP mechanics in
isolation — they're what composability across primitives unlocks.
Equal YES + NO holds zero outcome exposure.
A maker carrying balanced YES and NO of a fairly-priced outcome doesn't care which way it resolves — every pair pays out 1.00 USDH at settlement. They keep the spread, inventory is bounded by definition, and there's no funding to pay or liquidation to dodge.
e.g. delta-neutral two-sided quoting
Systematic short-vol, on-chain.
Recurring priceBinaries — auto-respawned BTC / ETH binaries every 15 minutes, hour, or day — are mechanically a stream of cash-secured short options. Strategies that trade premium in CME or Deribit translate directly, but with deterministic L1 settlement instead of a clearinghouse.
e.g. weekly BTC>$X programmatic underwriting
Hedges that span perps and outcome books.
HyperCore is one account system. If outcome books share margin with perps on mainnet — as the testnet design suggests — a perp vault can carry tail-risk insurance directly in outcome books, and an outcome-LP strategy can hedge directional drift in spot. That's composability you can't replicate by stitching together separate venues.
e.g. cross-primitive hedge construction
None of the above is shipped. HIP-4 is testnet-only, the spec hasn't been
published, and Compost's vault doesn't trade outcome books today. We're tracking it because the
same primitive that makes HIP-3 worth a capital formation layer makes HIP-4 worth one too — and
because the unified-margin angle only really works if someone's there with capital when mainnet
ships.
Sources
Where the numbers and mechanics come from.
HIP-4 has no official spec yet. The mechanics on this page are reverse-engineered from on-chain
transactions and the HyperCore API. Credit to the public researchers who did the legwork.
Primary research
Liquid Terminal · HIP-4 docs
The most thorough public reverse-engineering of HIP-4 — lifecycle, info endpoint, pair minting, system-wallet trace, and a clean industry comparison. liquidterminal.xyz/hip4/home ↗
Live data
Hyperliquid testnet API
POST api.hyperliquid-testnet.xyz/info with outcomeMeta and spotMetaAndAssetCtxs. Official docs ↗
Context
HIP-3 live board
For the perp side of the same engine — every builder market, real volume, no narrative. /hip3.html →