HIP-4 research update.
Correcting an assumption on the EVM contracts, new findings on multi-outcome mechanics, and next steps.
We got something wrong.
In our earlier analysis, we assumed the parimutuel contracts on HyperEVM (V1: 0x4fd7...44bd, V2: 0x6d86...cf07) were deployed by the Hyperliquid team. We had reasons to believe this:
The deployer wallet (0xe21c...d135) is over 2 years old with 19K transactions. The contracts were flagged as genesis deployments, but we've since learned that any contract on HyperEVM gets flagged this way. The timing correlated perfectly with the HIP-4 testnet launch, and the contracts are still actively used today with continuous contest creation, deposits, settlements, and fee withdrawals.
But after tracing the funding chain, the wallet that funded the deployer appears to be Block Theory Cap, an early HL user/fund, not the Hyperliquid team. No team members received genesis distributions. This is still a theory, not confirmed, but it changes the picture significantly.
If true, it means:
The official HIP-4 system is 100% HyperCore L1. Creation via registerTokensAndStandaloneOutcome, native pair minting (YES NO = 1.00), settlement via VoteGlobalAction. No EVM involved at any step.
The EVM parimutuel contracts are an independent project piggybacking on the same market data. Someone saw HIP-4 launch and immediately deployed their own parimutuel system on HyperEVM. Which is honestly impressive either way.
It's a total mystery right now. The timing, the activity level, the sophistication of the contracts (V2 has full OpenZeppelin security stack). Whether it's the team testing under a separate identity, a fund building on top of HIP-4, or something else entirely, we don't know yet.
Other findings from today:
Multi-outcome markets (food market with Akami, Canned Tuna, Otoro) work as independent YES/NO pairs on the CLOB. Posting an order on Akami doesn't affect Canned Tuna or Otoro at all. Each outcome is its own isolated orderbook with native pair minting. How the losing sides' collateral gets redistributed across multiple outcomes at settlement is still an open question we're investigating.
For off-chain data (who won the race, what did Hypurr eat),
@androolloyd pointed out that HIP-3's data feed model could serve as a base for pushing custom resolution data to HyperCore. The infrastructure exists. For subjective events it's obviously more complex than price feeds, but the mechanism is there.
What's next:
We keep digging.
@EnigmaValidator is joining the investigation and will help trace prediction market deployments on HyperCore, specifically multi-outcome creation transactions and the upcoming HIP-4 endpoints.
The game is far from over. We've been at this for 3 days straight and every layer we peel back reveals another question.
Stay tuned.
@androolloyd @0xExoMonk @EnigmaValidator