A friend texted me a screenshot and asked if I had seen the new market on Hyperliquid for a pre-IPO company’s synthetic price, I hadn’t, my first thought was that listing something like that usually takes a legal team, a data vendor contract, and months of chasing emails nobody answers, this one just appeared overnight with tight spreads and instant fills, like whoever built it hadn’t gotten the memo that it’s supposed to be hard.
That screenshot cost me my whole morning, by the time I got through the Hyperliquid docs, and few Twitter threads I had my answer: It’s HIP-3
What Is HIP-3?
For most of crypto’s history, an exchange did everything itself, it built the matching engine, picked which assets to list, set the risk parameters, and sourced its own price feeds, if you were a developer with a decent idea for a niche index or a regional commodity, you had no path forward unless you were willing to build an entire exchange around it, Hyperliquid Improvement Proposal 3 breaks that bundle apart.
HyperCore, Hyperliquid’s execution engine, keeps doing what it already does well: matching orders, calculating margin, running liquidations, and settling trades, HIP-3 opens up a separate lane on top of that.
A developer, referred to as a deployer, puts up 500,000 HYPE as collateral and gets to launch their own perp market on HyperCore, they set the leverage limits, the fee structure, and which asset gets listed. HyperCore just keeps running the engine underneath.
Go back to that pre-IPO market for a second, someone wanted to build a perp around an asset no centralized exchange would touch without a year of due diligence.
Before HIP-3, that meant building your own order book, your own liquidation logic, your own everything, before you even got to the part where you pick the asset, now that person stakes their HYPE, sets a few parameters, and HyperCore does the rest. The asset itself isn’t really the interesting part, what’s interesting is that the exchange infrastructure stopped being something you had to build from scratch.
The Oracle Problem Nobody Talks About Until It Goes Wrong
Here’s the part that gets skipped over a lot, and it happens to be the part that matters most: where does the price come from?
Fast execution doesn’t help you if your mark price is wrong, a perpetual contract needs a live, accurate value for whatever it’s tracking to calculate funding and to know when to liquidate someone, if that number is off by even a couple seconds, someone with faster information will trade against it, and someone else will get liquidated.
This is the job Pyth Network ends up doing for HIP-3 deployers.
When someone launches a market on HIP-3, picking the oracle is on them, writing your own price feed from scratch for a brand new market is a bad idea, mostly because that’s exactly where things break. Pyth gives builders a way around that: low-latency, feeds pulled straight from market makers, trading firms, and exchanges that actually see the flow.
Pyth updates prices on-chain in well under a second, and it covers a lot more than crypto pairs, including equities, FX, and commodities, a HIP-3 deployer can just wire that data into their market instead of trying to build a pricing system from scratch.
Back to the pre-IPO example, funding rates only work if the reference price is actually current, if the deployer had used a lagging or thin feed, funding would have drifted, and someone paying attention would have traded that gap while other users got liquidated for no reason at all.
Sub-second Pyth data keeps that from happening, because liquidations trigger off where the asset is trading right now instead of where it was trading a few minutes ago (nobody notices this part when it works but everyone notices when it doesn’t).
Real Money on the Line
So what actually stops a deployer from listing garbage, or feeding their own market bad prices to wipe out users on purpose?
It’s Money! Nobody at Hyperliquid is manually approving each market, instead, that 500,000 HYPE stake sits there as collateral the whole time a market is live, and it can be slashed.
If a deployer runs a market maliciously, or does something that puts the protocol at risk, validators can vote to slash that stake, it stays locked and slashable even through the unstaking window, so there’s no quick way to grab the money and run once trouble starts, deployers earn fees for running honest markets and lose their stake for running dishonest ones.
That’s a strange thing to get used to if you’re coming from traditional finance, where you’re mostly trusting a name on a building, here you’re trusting that a specific person has half a million dollars worth of tokens they’ll lose the moment they try something.
Where This Goes
Building a working derivatives exchange used to take a lot of capital, a full engineering team, and months of building pricing infrastructure before you could even open for trading.
HIP-3 removes most of that, Pyth removes the rest, what’s left for a deployer to actually work on is the market itself, picking good assets, setting fair parameters, and building a reason for traders to show up.
References: