Hyperliquid Prediction Markets (2026): What HIP-4 Actually Does

Search results for hyperliquid prediction markets mostly describe a Polymarket rival that does not exist yet. What exists is HIP-4, and it is not a proposal waiting on a vote: it has a permanent page in the official Hyperliquid documentation next to HIP-1, HIP-2 and HIP-3, and it has been live on mainnet since early May 2026. Secondary coverage agrees on 2 May as the date, though we could not find a primary announcement fixing it.

Hyperliquid is a perpetual futures exchange first, and event contracts arrived there as a listing standard rather than as a new app. That framing decides almost everything readers want to know: what trades, what it costs, who can list a market, and whether you can use it at all.

What Hyperliquid Prediction Markets Actually Are

The HIP series is how Hyperliquid ships exchange primitives: HIP-1 for tokens, HIP-2 for hyperliquidity, HIP-3 for builder-deployed perps. HIP-4 is titled "Outcome markets," and the docs define the instrument in one line: outcomes are "fully collateralized contracts that settle within a fixed range." Fully collateralized means no leverage, since every position is backed by the maximum it can pay out. Fixed range means the payoff need not be a coin flip between $1.00 and zero, so the same standard covers prediction markets and bounded, option-like payoffs.

Readers arriving from a search engine expect the opposite: a front page of elections, sports and news markets, priced in cents, ready to trade. HIP-4 gives you the rails for that. Whether an interesting set of markets exists on those rails on any given day is a separate question, and the answer in August 2026 is that everything listed settles against a crypto price.

What HIP-4 Does Under the Hood

The most interesting piece of engineering is the merged order book. In the docs' words, "the order books of Yes and No tokens for the same outcome are merged to share liquidity," because buying Yes at a price of p is the same trade as selling No at 1 minus p. That equivalence is what the docs mean by "dual": the dual of an order is its mirror on the opposite token at the complementary price, and a dual balance is the opposite-side inventory the same position can be expressed as. A split book makes real depth look thinner than it is, so Hyperliquid does the merge in the matching engine, with a documented tiebreak. At the same merged price level, resting sell orders sort ahead of resting buy dual orders.

A few primitives sit around it. The negate and merge operations convert between primary and dual balances, which is how a holder redeems quote tokens before settlement. "Questions" link outcomes so that exactly one settles to Yes and the rest settle to No, which is how a multi-outcome race gets expressed. Settlement is generalized rather than strictly binary: Yes converts into settleFraction quote tokens and No into 1 minus that, with the fraction at 1 or 0 in the plain binary case. The spec deliberately names the collateral only as "quote tokens" and does not fix it to one asset, so read the quote asset off the specific market before you size a position.

The Launch, and What Actually Trades

Reported first-day activity was around 6.05 million contracts. That is a contract count and not a dollar figure: $1 is the maximum a contract can pay out, so roughly $6 million is a ceiling on notional rather than traded value, and we found no source publishing a first-day dollar volume. Cumulative outcome volume through early August 2026 has been reported at a few hundred million dollars, and we could not corroborate a precise figure against a primary tracker. Hyperliquid's perpetuals book has done about $5.0 trillion all time (DefiLlama, checked 10 August 2026), which puts outcome markets well under 0.01% of the exchange's own business.

Composition matters more than the total. At launch the docs described "a recurring binary outcome that settles daily at 06:00 UTC to the BTC mark price," and said multi-outcome markets "will be supported but are not part of the initial mainnet release." Three months on, the board has widened without changing character. Checked on 10 August 2026, the live set was separate daily binaries on BTC, ETH, SOL and HYPE plus a group of BTC contracts partitioning the closing price into buckets (below one level, between two levels, at or above the upper level), all settling at 06:00 UTC. Every one resolves against a HyperCore mark price. None is an election, a sports result or a news event. The board turns over every day, so check the current set before you trade. Predictefy's Q2 2026 volume report has the venue by venue context.

Hyperliquid Prediction Market Fees

Right now you pay nothing. The HIP-4 page states it in one line: "fees are currently zero for outcome markets for initial testing." Read that as a testing-phase parameter rather than a permanent property of the product, because the machinery for charging you is already written down in full.

That machinery has one genuinely unusual feature. Per the fee documentation, "outcome trading only charges fees when closing or settling, not when opening outcome positions," so entering will still cost nothing even once the rate moves off zero. Rebates do not exist: traders who would earn maker rebates on spot and perps instead pay zero on maker orders, and only fee-paying volume counts toward tiers. Builder codes, the tag a front end attaches to route itself a cut of your trade, work as they do on spot, so an interface built on HIP-4 can charge you while the exchange itself does not.

One thing to be clear about, because it is widely garbled: the trading fees page publishes no outcome rate at all, and the tier table on it is the perpetuals schedule. So there is one official rate, zero, and one official mechanism, and they are not in conflict. We checked both pages on 10 August 2026; re-check them before you trade, since the rate is the part designed to change. For a venue whose costs you can compute today, Kalshi charges takers 0.07 × contracts × price × (1 - price), rounded up per order, with a different coefficient on some series and a maker fee on designated markets. Our Kalshi fee breakdown works through the arithmetic.

Who Can List a Market, and Who Can Trade

Deployment is template-gated rather than open. The deployer documentation says validators "vote on outcome templates, which HIP-4 deployers use as the basis for permissionless deployments," so the object under governance is the template, not the individual listing. Once a template exists, a deployer creates outcomes from it without asking anyone, bounded by per-deployer limits on active outcomes and on outcomes deployed per day. The only published values are testnet ones: 10 active outcomes per deployer and 50 deployed per day. Mainnet limits are not in the docs. Deploying costs no gas, but active deployers must maintain a staking requirement for as long as they remain deployers, and the API exposes actions like registerStandaloneOutcomeFromTemplate, registerQuestionFromTemplate and settleOutcome.

In July 2026 Hyperliquid announced a permissionless deployer track. The proposed terms: stake 500,000 HYPE, locked for six months, a cap of 100 outcomes per deployer, and slashing by validator vote for unclear markets, wrong settlements, or resolutions delayed beyond a week. Price that stake at whatever HYPE trades for on the day you read this, since the dollar figure moves with the token and several write-ups quote a conversion that has since gone stale. As of August 2026 the rollout is testnet-first and we could not confirm it had reached mainnet. Date-stamp whatever you read elsewhere: coverage from before late July omits the announcement, and some coverage after it implies the change already shipped. One outlet reports 1 million HYPE, which looks like a mix-up with the HIP-3 threshold.

Trading access is the harder gate. Hyperliquid's Terms of Use classify US persons as Restricted Persons, and the interface is also closed to Ontario residents. This is a standing term rather than an August 2026 development, and the terms can be amended, so read the current restricted-jurisdiction list yourself rather than trusting a summary. Circumvention through a VPN or proxy is prohibited, and the pattern elsewhere is that detection works and frozen accounts hold real money, as our piece on where Polymarket is legal covers. On the policy side, the Hyperliquid Policy Center filed a joint comment letter with Multicoin Capital on 27 July 2026 supporting the CFTC's proposed prediction-market rules. That is an advocacy organisation rather than the exchange or its labs entity, and nothing in the filing is live. None of this is legal, tax or investment advice.

How It Differs From a Dedicated Venue

The structural difference is unified margin. Spot positions, perpetual contracts and outcome markets live in one account against the same collateral pool, so hedging a perp with an event contract means no moving funds between platforms and no posting collateral twice. No dedicated prediction market can offer that, because none of them run a perps book. Together with the merged order book, it makes a coherent case for building event contracts inside a derivatives exchange rather than beside one.

What you give up is everything a specialist provides: a wide market list, a resolution process built for messy real-world events, a US regulator, and a front end for people who do not already trade perps. Settling to a HyperCore mark price, the on-chain order book and pricing layer that Hyperliquid's spot and perps run on, is mechanically clean and needs no human judgment. The docs say nothing about how a subjective event would resolve, and nothing currently listed forces the question.

The Hyperliquid column says when fees are charged, not how much. The rate is zero during initial testing, and that is the cell most likely to change.

Where the Official Hyperliquid Prediction Market Link Points

There is no separate prediction markets site, which is exactly why people search for a link. Three pages in the official documentation at hyperliquid.gitbook.io are the primary sources for everything above: the specification at HIP-4: Outcome markets, the fee treatment on the trading fees page, and the builder surface at HIP-4 deployer actions.

Trading happens at app.hyperliquid.xyz, the same interface as perps and spot, because outcomes are listed on the same exchange rather than a separate one. Between the documentation and that app you have the whole official surface for HIP-4. Anything presenting itself as a dedicated Hyperliquid prediction markets site is not one, and high-value DeFi front ends attract convincing lookalikes, so reach the app from the documentation rather than from a sponsored search result. If you want to see what is listed on a given day without connecting a wallet, the community tracker at hip4markets.com maintains a live and historical outcome list; it is not run by Hyperliquid and should be treated as a third-party source. US persons will not get past the geofence at any of these addresses.

Frequently Asked Questions

Conclusion

HIP-4 is a serious piece of exchange engineering attached to a small and very literal market list, so treating it as a Polymarket competitor gets both halves wrong. The merged book and the settlement primitives reward a close read if you are building on them. What is actually listed in August 2026 is daily crypto-price outcomes settling to mark prices at 06:00 UTC, the fee rate is zero because the documentation calls it a testing parameter, and US persons cannot get in at all. Two things will move that picture: the permissionless deployer track, still testnet-first at the time of writing, and the moment zero stops being the rate. We will recheck this page when either does. Our guide to prediction market data sources compares venues on numbers.