When the market has an owner
Builder-deployed perps hand the parameters of a market to whoever deployed it. That is the whole opportunity and the whole risk, and it changes what a maker has to check before quoting.
What HIP-3 changes
On a normal exchange the venue defines the contract: what it tracks, what the fees are, how the oracle is built, what leverage is allowed. A builder-deployed market moves those decisions to whoever deployed it, while settlement and matching stay with the underlying chain.
For a maker that is a different situation from either a centralised venue or a plain DEX. The book behaves like the host chain's book, but the economics, and the rules, belong to a party whose incentives are their own.
- Matching and settlement: the host chain. Parameters: the deployer.
- Fees, oracle construction, leverage and listing are all deployer decisions.
- The rules can change, and the change is not announced by the exchange.
Why the spreads are worth looking at
New markets of this kind are frequently under-quoted simply because few makers have looked at them yet. A spread that would be arbitraged away in minutes on a major venue can persist for hours here, which is the entire commercial reason to be early.
That advantage is temporary by construction. It lasts as long as the market is obscure, and it ends without warning when someone larger arrives. Plan the position sizing on the assumption that today's spread is not next month's.
- Under-quoted books are the opportunity, and the opportunity has a clock on it.
- Measure the spread across a full day: obscure markets are wide at night and normal by day.
- Size for the spread you will have, not the one you found.
The checks that are specific to this venue type
Read the oracle construction before anything else. A mark price built from a thin source can move in ways that have nothing to do with the underlying, and your inventory is priced against it. Then check what the deployer can change and how quickly: a fee schedule that can be altered is a cost you do not control.
Finally, test the exits during a quiet hour before you need them in a loud one. On a market with few participants, the taker exit you are counting on may cost far more than the book suggests, because the depth on the other side is a handful of orders rather than a wall.
- Oracle first: your inventory is priced against it, not against the world.
- Find out what the deployer can change and how fast.
- Test the exit at a quiet moment, not at the moment you need it.
What we run there, and what we measured
We quote on Entropy, a HIP-3 venue on Hyperliquid, and the operational differences from a major book are real: fewer participants, wider spreads and a mark price that deserves more attention than usual. The strategy itself is the same one that works elsewhere: cap inventory, exit as a maker, requote often.
The lesson that transferred least well was parameter reuse. A recipe calibrated on one market moved to another and broke immediately, because the tick was worth a different number of basis points. That is not a HIP-3 problem specifically; it is just more visible when every market has different owners and different settings.
- Same discipline as anywhere: cap inventory, leave as a maker, requote often.
- Do not carry parameters across markets. Convert the tick to basis points first.
- Watch the mark price more closely than on a deep venue.
Questions
What is HIP-3?
A mechanism on Hyperliquid that lets a party other than the exchange deploy and operate a perpetual market. The deployer sets the parameters (fees, oracle, leverage, what the contract tracks) while matching and settlement remain with the chain.
Is market making on a builder-deployed perp riskier?
It carries a risk that does not exist on a normal venue: the parameters of the market belong to a party whose interests are their own and can change. In exchange, spreads are often wider because fewer makers are present.
How do I evaluate a HIP-3 market before quoting?
Read the oracle construction, find out what the deployer can change, measure the spread across a full day rather than a snapshot, and test an exit while the market is quiet. Then apply the same arithmetic as anywhere: round-trip cost against typical spread.
Can the same bot quote HIP-3 and regular perps?
The mechanism is the same, but the parameters must not be shared. The tick is worth a different number of basis points on each market, and a preset carried across without converting is the most common way to lose money quietly.