What is HIP-4?
HIP-4 is Hyperliquid’s native prediction market protocol. It turns a question with a defined settlement moment into tradable outcome tokens that live on the HyperCore spot order book alongside every other spot asset, sharing the same account and the same collateral as perpetual futures. That single design choice is what separates it from prediction markets built as standalone platforms: there is no second venue, no second balance and no bespoke matching engine, just an exchange market whose payoff happens to be a question.
This page explains how outcome tokens are priced, the three market shapes the protocol supports, who is allowed to settle a market and how that differs by tier, and where permissionless deployment currently stands. For the numbers, see HIP-4 stats.
How an outcome token works
Each side of an outcome has its own token. For a binary outcome, a price of $0.62 is commonly read as an implied 62% probability, not a guaranteed forecast. At binary settlement, one side receives one quote token per share and the other receives zero. Standalone outcomes can also settle fractionally, so read the specific payout terms.
Hyperliquid merges the Yes and No books for the same outcome: a Yes bid at price p corresponds to a No ask at 1 minus p. Split and merge operations convert between collateral and paired tokens. Separate outcomes inside a multi-outcome question still have separate books, linked by the question’s merge and negate operations.
Because these are order-book assets rather than positions in a platform ledger, you can sell before trading closes if liquidity is available. You can also post a limit order; a price shown on a chart does not guarantee a fill for your chosen order size.
The three market shapes
Single-strike price threshold
One asset, one price level, one settlement time: "will this asset be above this level at this moment". Two sides, Yes and No, that together pay out $1 at settlement. This is the most common shape on the protocol.
Multi-strike price range
The same underlying carved into bands by N thresholds, producing N+1 named outcomes plus a fallback for anything outside them. Exactly one band settles at $1; the rest settle at $0. Useful when you have a view on where a price lands rather than merely which side of one level it ends on.
Custom event
An event with written settlement terms, such as a sports fixture or policy decision. Check the specific market: protocol event markets and permissionless template markets can have different settlement authorities.
The shape is not a label we attach after the fact. It is encoded in the market’s own on-chain description as a structured field list, which is what makes HIP-4 markets machine-readable. A single-strike price market carries something of this form:
class:priceBinary|underlying:BTC|expiry:20260804-0600|targetPrice:62809|period:1dTemplated markets use a different grammar: a keyword-and-value list whose keywords are fixed by the approved template rather than by the protocol’s canonical class scheme. Reading the shape from the structured description rather than guessing from the market’s title is the difference between a page that keeps working when Hyperliquid ships a new template and one that quietly mislabels every market of that kind.
Who settles a market: the three trust tiers
“Who resolves this” is the question most explainers skip, and it is the one that decides whether a market is worth trading. HIP-4 has three answers, and every market declares which one applies to it.
Auto-settle (protocol)
Hyperliquid protocol code reads a price observation at the market’s expiry and settles from it. No human decision in the resolution path, and the expiry moment is published on-chain before anyone trades. The cheapest trust assumption of the three.
Validator vote (network)
The Hyperliquid validator set votes on the outcome of an off-chain event. This is the path for custom markets with a written resolution clause. The trust surface is the same one that secures bridge withdrawals on the network: no external oracle is introduced.
Deployer-settled (third party)
The market’s template and supplied values define its settlement terms. The deployer, or a sub-deployer it authorizes, submits the result for its venue. Standalone outcomes allow fractional settlement; question outcomes settle to zero or one. Read the terms and identify the settlement authority before trading.
Settlement follows the market’s terms
A price template may reference a perp mark, an index or another specified observation; a sports template describes a game result. The template and its supplied values define the terms. For permissionless markets, the deployer or an authorized sub-deployer submits the settlement action. That authorization is different from an open challenge process available to every trader.
Hyperliquid requires no active outcomes and completion of the minimum staking duration before a deployer can deactivate. Its template reference also describes malformed markets that violate a template’s semantic restrictions as slashable. This does not establish an automatic late-settlement penalty or guarantee the result of a dispute.
The practical consequence for a trader is simple. Before you take the other side of a deployer-settled market, look at who deployed it and how many of their markets are currently sitting past their settlement moment with nothing signed. That is public data, and it is on the deployers page.
Permissionless deployment: where it actually stands
Permissionless HIP-4 deployment is now live on Hyperliquid mainnet. Eligible third parties can register as deployers and launch markets from validator-approved templates. Each deployer is also the settlement authority for the markets they create.
The registry below is read live from Hyperliquid, so it shows approved templates and registered deployers without relying on a launch-era snapshot. New permissionless markets appear on Liquidiction automatically as deployers launch them.
Validator-approved template registry, read live from Hyperliquid mainnet
28 approved templates on Hyperliquid mainnet, with 4 registered deployers.
aiModelHeadToHead
{shortNameA} v {shortNameB}: {leaderboard} on {dateTime}
This market compares {modelA} ({shortNameA}) and {modelB} ({shortNameB}) on {dateTime} using {leaderboard} at {leaderboardUrl}. Benchmark version, category, filters, and evaluation settings are {benchmarkSettings}, applied identically to both models. Each model is the exact version and configuration identified by its leaderboard name. Renames count only if the leaderboard publisher identifies the same model; replacements and other configurations do not. A relocated or renamed leaderboard counts only if its publisher identifies the same evaluation with the specified settings. The Snapshot is the latest official leaderboard version matching those settings public at or before {dateTime}. Proof requires official publication metadata identifying that version or a verifiable timestamped capture of its contents by then. A date alone suffices only if the entire date precedes the deadline in an established timezone. Later captures qualify only if official metadata or earlier captures prove identical contents public by the deadline. Official version order prevails; capture order alone does not order versions. Unresolved conflicts over version selection or required facts leave them unestablished. Comparison favors the {comparisonDirection} value of {metric}, where direction is higher or lower. Tiebreakers are {tiebreakers}, with each metric and direction specified; none means no tiebreakers. They apply in the specified order only after equality on all earlier criteria. Only specified criteria apply, using displayed values at published precision. Ambiguous settings or comparison criteria leave the comparison unestablished. The market resolves to 0.5 if either model is absent, both names identify the same model, the Snapshot, comparison, or a required fact is unestablished by the evidence cutoff, or the leaderboard was discontinued without a qualifying successor by {dateTime}. Otherwise, it resolves to {shortNameA} if comparison favors {modelA}, to {shortNameB} if it favors {modelB}, and to 0.5 if tied after all specified criteria. A resolution to 0.5 pays $0.50 per token on either side. Unused tiebreak values are unnecessary. Evidence is limited to the leaderboard publisher's official records and verifiable timestamped leaderboard captures public by 24 hours after {dateTime}; resolution occurs only after that cutoff. Evidence may authenticate the Snapshot as published, including model identities; corrections, reassignments, and versions first public after {dateTime} are excluded even if their data covers an earlier period. All comparisons use the same Snapshot. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:technology|subCategory:N/A|semanticRestriction:This market must compare two specific AI language models on a public performance leaderboard.
- benchmarkSettings: string
- comparisonDirection: string
- dateTime: dateTime
- leaderboard: string
- leaderboardUrl: string
- metric: string
- modelA: string
- modelB: string
- shortNameA: shortString
- shortNameB: shortString
- tiebreakers: string
Sides: {shortNameA} / {shortNameB}
awardCandidate
{candidate}
This outcome resolves to Yes if, under the parent question's rules, it is the earliest registered candidate not otherwise resolved to No that identifies the one winning entry of the Award, and otherwise resolves to No. Its Subject is {subject}. A Subject that identifies a work, group, or named person on more than one entry, or that {candidate} does not name in full or in short form, identifies no entry. Resolution follows the parent question's timing and source rules. {candidate} is a display label.
- candidate: string
- subject: string
awardEntryWins
{displayName} wins {period} {award}?
This {field} market covers whether {entry} wins the {period} {award} (the "Award"). {entry} identifies, in the Award's context, one person, group, organization, or work and need not be a nominee. {entry} wins if official channels have named it as of fixing, under any name unambiguously identifying it, as a winner, recipient, laureate, fellow, or honoree of the Award, including as a named recipient for a winning work, team, or group, or as the work for which a winner is named; being credited on, a member of, affiliated with, or accepting for a winning entry without being so named does not count. A result is complete once official channels have announced every winner of the Award; nominations, shortlists, and partial announcements do not complete it. Official channels are the publications, events, and broadcasts of {awardingBody} and the Award's conferring body and administrator. The trigger is the earliest verifiable announcement on an official channel that establishes a Yes or No condition below; an announcement is not a trigger if, within one hour after it and before {resolutionDeadline}, it is withdrawn or removed or an official channel contradicts it. Fixing is the earlier of one hour after the trigger and {resolutionDeadline}, or {resolutionDeadline} without a trigger. Facts are as official channels have stated them as of fixing; a verifiably later statement supersedes earlier ones on the same channel or made at an event or broadcast, an explicit correction by fixing prevails across channels, other conflicts leave a fact unestablished, and omission is not a conflict. At fixing, the market resolves to 0.5 if {entry} names a class, alternatives, or more than one entry, or a winner could be {entry} but {entry} does not identify exactly one entry; if {displayName} is not a name or short form of {entry}; if, among editions of one award that {awardingBody} confers or administers with a winner due by {resolutionDeadline}, {period} does not identify one by {awardingBody}'s labeling or, failing that, by the year its winner is due; or if neither condition below is established, including because the result is incomplete. Otherwise, it resolves to Yes if {entry} wins, even if other entries also win, and to No if the result is complete and {entry} does not win, or official channels announce the Award is withheld or will not be conferred; postponement or reservation is not a No condition. Only evidence public by 24 hours after fixing establishes the trigger and the facts at fixing; timestamps must prove when contents were public; resolution occurs only after that cutoff. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:culture|subCategory:{field}|semanticRestriction:This market must refer to whether an entry is among the winners of an award or honor, not a selection for office or candidacy.
- award: string
- awardingBody: string
- displayName: string
- entry: string
- field: string
- period: string
- resolutionDeadline: dateTime
Sides: Yes / No
awardOther
Other
This outcome resolves to Yes if the parent question's Other conditions apply, and otherwise resolves to No.
awardWinner
{period} {award} winner
This {field} market covers the {period} {award} (the "Award"). Among editions of one award that {awardingBody} confers or administers with a winner due by {resolutionDeadline}, {period} identifies one by {awardingBody}'s labeling or, failing that, by the year its winner is due; otherwise the Award Result is unestablished. It has one named outcome per listed candidate plus one called Other, which may be registered at any time. Exactly one outcome resolves to Yes; all others resolve to No. An entry is one nominee, finalist, or winner of the Award as official channels record it, including a group recorded as one, with every work and person named on it as its nominee or recipient, even jointly, not as a member or representative of a group. A candidate identifies an entry if its Subject unambiguously identifies a work, group, or named person on that entry and no other entry of the Award. Official channels are the publications, events, and broadcasts of {awardingBody} and the Award's conferring body and administrator. The trigger is the earliest verifiable announcement on an official channel of the result: a winner, winners, or that the Award is withheld or not conferred; an announcement is not a trigger if, within one hour after it and before {resolutionDeadline}, it is withdrawn or removed or an official channel contradicts it. Fixing is the earlier of one hour after the trigger and {resolutionDeadline}, or {resolutionDeadline} without a trigger. The Award Result is the winner official channels have named as of fixing; a verifiably later statement supersedes earlier ones on the same channel or made at an event or broadcast, an explicit correction by fixing prevails across channels, other conflicts leave it unestablished, and omission is not a conflict. A candidate registered after the earliest official-channel announcement of a winner of the Award, whether or not it is a trigger, or that identifies no entry at fixing, resolves to No. A candidate finally resolves to No once official channels publish final nominees or finalists none of which it identifies or show it identifies the same entry as an earlier registered candidate. Of the remaining candidates, the earliest registered identifying the one winning entry the Award Result names resolves to Yes; all others resolve to No. Other resolves to Yes if no candidate resolves to Yes, including if the Award Result is unestablished at fixing, names no winner, or names more than one winning entry, such as a tie. Only evidence public by 24 hours after fixing establishes the trigger and the Award Result at fixing; timestamps must prove when contents were public; a Yes resolution occurs only after that cutoff. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:culture|subCategory:{field}|semanticRestriction:This market must refer to the single winner of an award or honor, not a selection for office or candidacy.
- award: string
- awardingBody: string
- field: string
- period: string
- resolutionDeadline: dateTime
binaryPrice
{perp} above {threshold} at {time}?
The market resolves to Yes if the {perp} price is above {threshold} at {time}, and otherwise resolves to No. Settlement is according to the {seconds}-second TWAP of {priceDescription} price ending at {time}. If {perp} is delisted before this market settles, settlement is instead according to the {perp} settlement price at delisting. metadata=category:price|subCategory:N/A|semanticRestriction:This market must refer to the price of a Hyperliquid perp.
- perp: hlPerp
- priceDescription: string
- seconds: uInt
- threshold: uDecimal
- time: dateTime
Sides: Yes / No
binaryPriceExternal
{shortName} above {threshold} on {time}?
The market resolves to Yes if the {instrument} price is above {threshold} on {time}. Otherwise, the market resolves to No. Settlement is according to the {seconds}-second TWAP of {priceDescription} price from {feed} ending on {time}. If {feed}'s authority permanently stops publishing a qualifying price for {instrument} before that TWAP window ends, settlement is instead according to that authority's official final or delisting price for {instrument}, or, if it designates none, the last qualifying price {feed} published for {instrument}. If {feed} publishes no qualifying price for {instrument} within that window for any other reason, settlement is instead according to the last qualifying price {feed} published for {instrument} before the window ended. If no settlement price under these rules is publicly verifiable by 24 hours after {time}, the market instead resolves to 0.5, and evidence first public after that cutoff does not affect resolution. metadata=category:price|subCategory:N/A|semanticRestriction:This market must refer to a publicly verifiable price from an authoritative external feed.
- feed: string
- instrument: string
- priceDescription: string
- seconds: uInt
- shortName: shortString
- threshold: uDecimal
- time: dateTime
Sides: Yes / No
companyIpoConfirmed
{company} IPO confirmed by {dateTime}
The market resolves to Yes if {company}'s first public common-equity offering for a regulated exchange listing is publicly confirmed at or before {dateTime}. Otherwise, it resolves to No. Confirmation occurs when any of these becomes public: a securities regulator approves or clears the required filing for the public offering; the company or underwriters publish final share count and offering price; or the exchange assigns a ticker and grants unconditional listing approval. Conditional exchange approval qualifies only when the exchange publicly confirms all conditions satisfied. These triggers are independent. All triggers require a qualifying offering. Filing or confidential submission, a proposed price range, ticker reservation, or intent alone do not qualify. Trading need not begin by the deadline. Issuer identity and offering structure are assessed when each trigger becomes public. Prior trading means trading before that trigger, excluding trading resulting from that offering. Later restructuring, delay, withdrawal, or cancellation does not undo qualifying confirmation; corrections may show it never qualified. Name changes of the same issuer count; offerings by a parent, subsidiary, or acquirer do not. Prior public common-equity trading on a regulated exchange disqualifies the company, even if later taken private. Prior trading outside a regulated exchange disqualifies it if retail investors could buy and sell its common equity without investor qualification requirements, invitation, or issuer approval. Routine account opening and identity checks do not restrict access. A direct listing providing for newly issued common shares to be sold to the public qualifies; a resale-only direct listing does not. SPAC transactions, reverse mergers, private placements, and offerings of debt, preferred shares, or depositary receipts without a common-equity IPO do not qualify. Evidence and corrections must be public by seven days after {dateTime}; resolution occurs only after that cutoff. They establish whether a qualifying confirmation was public by the deadline. Official source priority is regulator, company, exchange, then underwriters. Priority resolves contradictions about the same fact; silence does not negate another source's confirmation. Unresolved conflicts at the highest applicable priority leave a fact unestablished. Otherwise, at least two independent reputable news organizations' original reports may establish missing facts if reputable reporting has no unresolved material contradiction. Syndicated copies count as one. Reporting cannot override official facts or resolve official conflicts; rumors do not qualify. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:economics|subCategory:N/A|semanticRestriction:This market must refer to confirmation of a company's initial public offering.
- company: string
- dateTime: dateTime
Sides: Yes / No
companyIpoFirstDayMarketCap
{company} first-day market cap above ${marketCapThresholdB}B?
This market covers {company}'s first public offering of common equity for a regulated exchange listing (the "Offering"). The Offering does not qualify if {company}'s common equity traded publicly before the Offering on a regulated exchange or on a venue open to retail investors without qualification, or if it is a direct listing, SPAC transaction, reverse merger, or offers units, debt, preferred shares, or depositary receipts not representing only common equity. Renames and a holding company formed for the Offering count as {company}; another parent, subsidiary, or acquirer does not. The Final Prospectus is the final prospectus or equivalent filed with the securities regulator for the Offering, with supplements filed by the evidence cutoff; the Listing Exchange is the regulated exchange it names for the listing or, if several, the first named that quotes a pricing currency of the Offering, else the first named. The First Trading Day is the first day a trade in the offered securities executes in the Listing Exchange's regular session, excluding conditional or when-issued dealings. The Closing Price is the Listing Exchange's official closing price for them that day; for depositary receipts, divided by the shares each represents. Shares Outstanding is all common or ordinary shares of every class outstanding immediately after the Offering at its final size excluding over-allotment shares, as the Final Prospectus states or its stated figures determine; for a class with several figures on that basis, the first in the prospectus summary applies. The Market Cap is the Closing Price times Shares Outstanding, in US dollars. A Closing Price in another currency converts at the International Monetary Fund's representative rate or, for a currency it omits, the cross rate from the European Central Bank's euro reference rates, or, for one both omit, its central bank's reference rate, for the First Trading Day or, if none, the latest earlier day; otherwise the Market Cap is unestablished. The market resolves to 0.5 if {company} does not identify a single company, the Offering does not qualify, no First Trading Day's regular session has closed by {listingDeadline}, or a required fact is unestablished by the evidence cutoff. Otherwise, it resolves to Yes if the Market Cap exceeds {marketCapThresholdB} billion and to No if not. Only evidence public by seven days after the earlier of the First Trading Day's close and {listingDeadline} is used; resolution occurs only after that cutoff. A correction a source publishes by the cutoff prevails; conflicts these rules do not resolve leave a fact unestablished. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:economics|subCategory:N/A|semanticRestriction:This market must refer to the first-day closing market capitalization of a company's initial public offering.
- company: string
- listingDeadline: dateTime
- marketCapThresholdB: uDecimal
Sides: Yes / No
policyRateBucket
{bucketLabel}
This outcome resolves to Yes if the Change defined by the parent question is at least {lowerBps} basis points and less than {upperBps} basis points, and none of the parent question's fallback conditions applies. Otherwise, it resolves to No. Resolution follows the parent question's timing and source rules. The lower bound is inclusive and the upper bound is exclusive. Finite bounds use decimal notation in basis points with an optional leading sign and no units, grouping separators, or exponent. "-inf" is permitted only as the lower bound and "+inf" only as the upper bound; these mean no lower or upper bound, respectively. The lower bound must be strictly less than the upper bound. All listed ranges must be disjoint. {bucketLabel} is a display label; the bounds determine the range.
- bucketLabel: string
- lowerBps: string
- upperBps: string
policyRateDecision
{institution} {decisionLabel} rate decision
This market has three possible results: No change, Decrease, or Increase. Exactly one corresponding named outcome resolves to Yes; the other two and the protocol-created fallback resolve to No. It covers the {decisionLabel} {institution} decision concerning {policyMeasure}, scheduled for {scheduledDecision} UTC (the "Decision"). The Decision Value is the numeric value of {policyMeasure} announced by {officialSource}. The Prior Value is the value of that same measure in effect immediately before the Decision. No change resolves to Yes if the Decision Value equals the Prior Value. Decrease resolves to Yes if the Decision Value is lower. Increase resolves to Yes if the Decision Value is higher. If the Decision is rescheduled, the rescheduled decision applies if {officialSource} publishes it by {decisionDeadline} UTC. If the Decision is canceled or postponed without a replacement decision, or no official Decision Value is published by {decisionDeadline} UTC, No change resolves to Yes. Policy changes outside the scheduled Decision are ignored except when establishing the Prior Value. If {institution} replaces {policyMeasure}, the closest official successor measure identified by {officialSource} is used. Resolution uses the initially published decision; later revisions do not change the resolution. {officialSource} is the primary resolution source; independent reputable reporting may be used if it has not published the relevant decision. No change resolves to Yes under the deadline rule above only if neither {officialSource} nor independent reputable reporting establishes a Decision Value by {decisionDeadline} UTC. metadata=category:economics|subCategory:N/A|semanticRestriction:This market must refer to a scheduled policy rate decision by an official institution.
- decisionDeadline: dateTime
- decisionLabel: string
- institution: string
- officialSource: string
- policyMeasure: string
- scheduledDecision: dateTime
policyRateDecrease
Decrease
This outcome resolves to Yes if the Decision Value is lower than the Prior Value under the parent question's rules.
policyRateIncrease
Increase
This outcome resolves to Yes if the Decision Value is higher than the Prior Value under the parent question's rules.
policyRateLadder
{institution} {decisionLabel} rate decision
This market covers the {decisionLabel} {institution} decision concerning {policyMeasure}, scheduled on {scheduledDecision} (the "Decision"). It has one named outcome per listed change bucket plus the fallback. Exactly one outcome resolves to Yes; all others resolve to No. The Decision Value is the value announced for {policyMeasure} in the first official announcement of the Decision, even if effective later. If it explicitly leaves the measure unchanged without restating a value, the Decision Value equals the Prior Value. The Prior Value is the value of the same measure in effect immediately before that announcement, including intervening unscheduled changes. The Change is the Decision Value minus the Prior Value, in basis points: one basis point is 0.01 percentage point. Values use published precision; derived values and the Change are calculated without rounding. {policyMeasure} identifies one finite numeric rate or a specified bound or midpoint of a target range. Both values use the same definition and units. Ambiguous measures or values leave the Change unestablished. If the measure is replaced, a successor counts only if {officialSource} explicitly identifies it and establishes both values on the same basis; otherwise the Change is unestablished. A rescheduled or replacement decision counts only if {officialSource} identifies it as replacing the Decision. Its announcement must be public by {decisionDeadline}. Other decisions affect only the Prior Value. Cancellation or postponement alone does not establish zero Change. Only evidence public by {decisionDeadline} is used; resolution occurs after that deadline. {officialSource}'s official records take priority. If a required fact is missing, at least two independent reputable news organizations' original reports may establish it, with no unresolved material contradiction in reputable reporting. Syndicated copies count as one. Reporting must describe an official announcement, cannot override official facts, and cannot resolve official conflicts. Unresolved official conflicts leave the affected fact unestablished. Only the initially announced decision applies; later official revisions and corrections to it are ignored. Each bucket includes its lower bound and excludes its upper bound. All listed ranges must have valid bounds as defined by their outcomes and be disjoint; labels do not alter their bounds. Unless a fallback condition applies, the named outcome whose range contains the Change resolves to Yes and all other outcomes resolve to No. The fallback resolves to Yes and every named outcome resolves to No if neither the Decision nor a qualifying replacement is announced by {decisionDeadline}, either value or the Change is unestablished by then, any range is invalid, ranges overlap, or no range contains the Change. metadata=category:economics|subCategory:N/A|semanticRestriction:This market must refer to a scheduled policy rate decision by an official institution.
- decisionDeadline: dateTime
- decisionLabel: string
- institution: string
- officialSource: string
- policyMeasure: string
- scheduledDecision: dateTime
policyRateNoChange
No change
This outcome resolves to Yes if the Decision Value equals the Prior Value, or if the parent question's cancellation, postponement, or decision-deadline rule applies.
priceTouch
{perp} touches {target} by {time}
The market resolves to Yes if the {perp} price touches {target} at or before {time}, and otherwise resolves to No. The price counts as touched once the {seconds}-second TWAP of {priceDescription} price equals or crosses {target} at any time after this market is deployed and at or before {time}; the market can settle as soon as a touch occurs. If {perp} is delisted before this market settles, the {perp} settlement price at delisting counts as the final price for the touch determination, and the market can settle at delisting if the price has not touched {target} by then. metadata=category:price|subCategory:N/A|semanticRestriction:This market must refer to the price of a Hyperliquid perp.
- perp: hlPerp
- priceDescription: string
- seconds: uInt
- target: uDecimal
- time: dateTime
Sides: Yes / No
scalarPrice
{perp} from {low} to {high} at {time}
If the {perp} price is at or above {high} at {time}, Long tokens pay out $1 each. If the price is at or below {low}, Long tokens pay out $0 each. Otherwise, Long tokens pay out $1 times (price - {low}) / ({high} - {low}) each. In each case, Short tokens pay out $1 minus the Long payout each. Settlement is according to the {seconds}-second TWAP of {priceDescription} price ending at {time}. If {perp} is delisted before this market settles, settlement is instead according to the {perp} settlement price at delisting. metadata=category:price|subCategory:N/A|semanticRestriction:This market must refer to the price of a Hyperliquid perp.
- high: uDecimal
- low: uDecimal
- perp: hlPerp
- priceDescription: string
- seconds: uInt
- time: dateTime
Sides: Long / Short
sportsContestDraw2
Draw
This outcome resolves to Yes if, under the parent question's rules, the Contest Result is a draw or the Residual Conditions apply, and otherwise resolves to No.
sportsContestParticipant2
{participant}
This outcome resolves to Yes if {participant} is the sole winner of the Contest under the parent question's rules, and otherwise resolves to No.
- participant: string
sportsContestResult
{competition} {stage}: {participantA} v {participantB}
This {sport} market has three possible results: {participantA}, Draw, or {participantB}. Exactly one corresponding named outcome resolves to Yes; the other two and the fallback resolve to No. It covers the {contestType} between {participantA} and {participantB} in the {season} {competition} {stage}, scheduled for {scheduledStart} UTC (the "Contest"). The Counted Play is {countedPlay}. The Contest Result is the official record of the Counted Play. Play outside the Counted Play, including any extra time, overtime, shoot-out, or tiebreak, is disregarded, and a winner decided by such play is not the winner of the Counted Play, even if any source treats it as part of the Contest. A participant is the sole winner if the Contest Result names it the sole winner of the Counted Play or, where it names none, if it has the better score or placing under the Contest rules when the Counted Play ends. The Contest Result is a draw if it records a draw, tie, or co-winners, or names no winner of the Counted Play with the scores level when it ends. {participantA} or {participantB} resolves to Yes if that participant is the sole winner; Draw resolves to Yes if the Contest Result is a draw. Draw also resolves to Yes if the Contest Result records a no contest or no winner, names an unlisted winner, or is not established by {resolutionDeadline} UTC (the "Residual Conditions"). A result that {officialSource} records for the Contest by administrative or rules-based decision, including but not limited to a decision following abandonment, walkover, forfeit, disqualification, or retirement, is the Contest Result if the Counted Play was not completed; a completed Counted Play is resolved on its record alone. A postponed, suspended, or rescheduled Contest, or one replayed because the Counted Play was not completed, between the same participants in the same {competition} {stage} remains the Contest. A result or status is established when {officialSource} publishes it or, failing that by {resolutionDeadline} UTC, when a clear consensus of independent reputable reporting available by that deadline supports it; otherwise it is not established. The Contest Result is fixed as {officialSource}'s record stands one hour after {officialSource} first publishes on any official channel that the Counted Play is complete, including by recording play beyond it, or that a result has been recorded by decision, or at {resolutionDeadline} UTC if earlier or if it rests on reporting; later appeals, corrections, reversals, or reassignments do not change it. This market is not sponsored by, endorsed by, or affiliated with {officialSource} or any competition, league, governing body, organizer, team, club, participant, or other person or entity referenced in this market. All names and trademarks are the property of the respective rights holders and are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to a single scheduled sports contest.
- competition: string
- contestType: string
- countedPlay: string
- officialSource: string
- participantA: string
- participantB: string
- resolutionDeadline: dateTime
- scheduledStart: dateTime
- season: string
- sport: string
- stage: string
sportsContestWinner
{competition} {stage}: {participantA} v {participantB}
This {sport} market has two sides, {shortNameA} for {participantA} and {shortNameB} for {participantB}. It covers the {contestType} between {participantA} and {participantB} in the {season} {competition} {stage}, scheduled for {scheduledStart} UTC (the "Contest"). The Counted Play is {countedPlay}. The Contest Result is the official record of the Counted Play. Play outside the Counted Play, including any extra time, overtime, shoot-out, or tiebreak, is disregarded, and a winner decided by such play is not the winner of the Counted Play, even if any source treats it as part of the Contest. A participant is the sole winner if the Contest Result names it the sole winner of the Counted Play or, where it names none, if it has the better score or placing under the Contest rules when the Counted Play ends. The Contest Result is a draw if it records a draw, tie, or co-winners, or names no winner of the Counted Play with the scores level when it ends. The market resolves to {shortNameA} if {participantA} is the sole winner, and resolves to {shortNameB} if {participantB} is the sole winner. The market resolves to 0.5 if the Contest Result is a draw, records a no contest or no winner, names an unlisted winner, or is not established by {resolutionDeadline} UTC. A result that {officialSource} records for the Contest by administrative or rules-based decision, including but not limited to a decision following abandonment, walkover, forfeit, disqualification, or retirement, is the Contest Result if the Counted Play was not completed; a completed Counted Play is resolved on its record alone. A postponed, suspended, or rescheduled Contest, or one replayed because the Counted Play was not completed, between the same participants in the same {competition} {stage} remains the Contest, subject to the same rules and deadline. A result or status is established when {officialSource} publishes it or, failing that by {resolutionDeadline} UTC, when a clear consensus of independent reputable reporting available by that deadline supports it; otherwise it is not established. The Contest Result is fixed as {officialSource}'s record stands one hour after {officialSource} first publishes on any official channel that the Counted Play is complete, including by recording play beyond it, or that a result has been recorded by decision, or at {resolutionDeadline} UTC if earlier or if it rests on reporting; later appeals, corrections, reversals, or reassignments do not change it. This market is not sponsored by, endorsed by, or affiliated with {officialSource} or any competition, league, governing body, organizer, team, club, participant, or other person or entity referenced in this market. All names and trademarks are the property of the respective rights holders and are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to a single scheduled sports contest.
- competition: string
- contestType: string
- countedPlay: string
- officialSource: string
- participantA: string
- participantB: string
- resolutionDeadline: dateTime
- scheduledStart: dateTime
- season: string
- shortNameA: shortString
- shortNameB: shortString
- sport: string
- stage: string
Sides: {shortNameA} / {shortNameB}
sportsOverUnderMarket
{event}: {measure} over/under {line}
This {sport} market covers {measure} in {event}, scheduled for {scheduledStart} (the "Contest"). Counted Play is {countedPlay}; extra time, overtime, shoot-outs, and tiebreaks count only if expressly included. The Final Value uses only Counted Play. It and {line} use {measure}'s units and scale. Reported values use published precision. Derived measures, including totals, differences, ratios, and averages, are calculated from underlying statistics without rounding. Unclear units or calculations, undefined values, or zero denominators leave the Final Value unestablished. Completion requires all specified play to conclude. For full-contest scopes, normal endings under rules in force at scheduled start, including mercy rules and unneeded periods, count. Abandonment, retirement, weather-shortening, forfeits, and awarded results do not complete unfinished Counted Play. Postponement, suspension, rescheduling, or replay after incomplete Counted Play retains the named teams or competitors and deadline; team lineups may change. Resumed play continues the record; a full replay replaces it. An individual participates only by entering Counted Play. The market resolves to 0.5 if completion, the Final Value, or participation by every individual to whom {measure} refers is unestablished at fixing. Otherwise, it resolves to Over if the Final Value exceeds {line}, to Under if below, and to 0.5 if equal. A resolution to 0.5 pays $0.50 per token on either side. Only evidence public by {resolutionDeadline} establishes the trigger time and source contents at fixing; resolution occurs only after that deadline. The trigger is the earliest verifiable official publication time for a record establishing completion, final statistics, and whether each relevant individual participated; absent that, its earliest verifiable capture time. Timestamps must prove those contents existed then. Fixing is the earlier of one hour after the trigger and {resolutionDeadline}, or that deadline without a trigger. {officialSource}'s record establishes the facts. A verifiably newer version of the same record supersedes older versions. Across official channels, an explicit correction or designation of authority prevails; other conflicts leave the fact unestablished. Only versions public by fixing determine facts. Only if fixing occurs on {resolutionDeadline} may missing facts be established by at least two independent reputable news organizations' original reports, with no unresolved material contradiction in reputable reporting. Syndicated copies count as one. Reporting cannot override official facts or resolve official conflicts. Changes after fixing and evidence after the deadline do not affect resolution. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to a numerical measure from a single scheduled sports contest.
- countedPlay: string
- event: string
- line: uDecimal
- measure: string
- officialSource: string
- resolutionDeadline: dateTime
- scheduledStart: dateTime
- sport: string
Sides: Over / Under
sportsScalarMarket
{event}: {measure} from {low} to {high}
This {sport} market has two sides, Over and Under. It covers {measure} in the {event}, scheduled for {scheduledStart} UTC (the "Contest"). The Counted Play is {countedPlay}. The Final Value is the value of {measure} over the Counted Play only, taken from the official record of the Contest. Play outside the Counted Play, including any extra time, overtime, shoot-out, or tiebreak, is disregarded even if {officialSource} or any other source includes it in the Contest or in its published totals. Where {officialSource} publishes no total scoped to the Counted Play, the Final Value is determined from the period-level or event-level data in its official record. If the Final Value is at or above {high}, Over tokens pay out $1 each. If the Final Value is at or below {low}, Over tokens pay out $0 each. Otherwise, Over tokens pay out $1 times (Final Value - {low}) / ({high} - {low}) each, rounded down to the nearest $0.00000001. In each case, Under tokens pay out $1 minus the Over payout each. A postponed, suspended, or rescheduled Contest, or one replayed because the Counted Play was not completed, remains the Contest, subject to the same rules and deadline. If the Contest is shortened, abandoned, or decided by administrative or rules-based decision but {officialSource} publishes an official final value of {measure} for the Contest, that value, counting only the Counted Play, is the Final Value. A participant takes part in the Counted Play if {officialSource}'s record shows it in play at any point during the Counted Play. The market resolves to 0.5 if {low} is not less than {high}, if any individual participant to whom {measure} refers does not take part in the Counted Play, if the Final Value cannot be determined from the official record, or if no Final Value is established by {resolutionDeadline} UTC. A value or status is established when {officialSource} publishes it or, failing that by {resolutionDeadline} UTC, when a clear consensus of independent reputable reporting available by that deadline supports it; otherwise it is not established. The Final Value is fixed as {officialSource}'s record stands one hour after {officialSource} first publishes on any official channel that the Counted Play is complete, including by recording play beyond it, or that a value has been recorded by decision, or at {resolutionDeadline} UTC if earlier or if it rests on reporting; later appeals, corrections, reversals, or reassignments do not change it. This market is not sponsored by, endorsed by, or affiliated with {officialSource} or any competition, league, governing body, organizer, team, club, participant, or other person or entity referenced in this market. All names and trademarks are the property of the respective rights holders and are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to an officially reported statistic from a single scheduled sports contest.
- countedPlay: string
- event: string
- high: uDecimal
- low: uDecimal
- measure: string
- officialSource: string
- resolutionDeadline: dateTime
- scheduledStart: dateTime
- sport: string
Sides: Over / Under
sportsSpread
{participantA} v {participantB}: {shortNameA} {spread}
This {sport} market covers the contest between {participantA} ({shortNameA}) and {participantB} scheduled for {scheduledStart} (the "Contest"). Counted Play is {countedPlay}; extra time, overtime, shoot-outs, and tiebreaks count only if expressly included. The Final Value is {participantA}'s {measure} plus {spread} minus {participantB}'s {measure}, each {measure} taken over Counted Play only. {spread} is a + or - sign followed by a decimal in {measure}'s units and scale. Reported values use published precision. Derived measures are calculated from underlying statistics without rounding. Unclear units or calculations leave the Final Value unestablished. Completion requires all Counted Play to conclude. For full-contest scopes, normal endings under rules in force at scheduled start, including mercy rules and unneeded periods, count. Abandonment, retirement, weather-shortening, forfeits, and awarded results do not complete unfinished Counted Play. Postponement, suspension, rescheduling, or replay after incomplete Counted Play retains the participants and deadline; lineups may change. Resumed play continues the record; a full replay replaces it. An individual participates only by entering Counted Play. The market resolves to 0.5 if completion, the Final Value, or any individual participant's participation is unestablished at fixing. Otherwise, it resolves to Yes if the Final Value exceeds 0, to No if below, and to 0.5 if equal. Only evidence public by {resolutionDeadline} establishes trigger time and source contents at fixing; resolution occurs only after that deadline. The trigger is the earliest verifiable official publication time for a record establishing completion, final statistics, and whether each relevant individual participated; absent that, its earliest verifiable capture time. Timestamps must prove contents existed then. Fixing is the earlier of one hour after the trigger and {resolutionDeadline}, or that deadline without a trigger. {officialSource}'s record establishes the facts. A verifiably newer version supersedes older versions of that record. Across official channels, an explicit correction or designation of authority prevails; other conflicts leave facts unestablished. Only versions public by fixing determine facts. Only if fixing is on {resolutionDeadline} may at least two independent reputable news organizations' original reports establish missing facts, with no unresolved material contradiction in reputable reporting. Syndicated copies count as one. Reporting cannot override official facts or resolve official conflicts. Changes after fixing and evidence after the deadline do not affect resolution. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to a spread in a single scheduled sports contest.
- countedPlay: string
- measure: string
- officialSource: string
- participantA: string
- participantB: string
- resolutionDeadline: dateTime
- scheduledStart: dateTime
- shortNameA: shortString
- sport: string
- spread: string
Sides: Yes / No
sportsTotal
{participantA} v {participantB}: total {line}
This {sport} market covers the contest between {participantA} and {participantB} scheduled for {scheduledStart} (the "Contest"). Counted Play is {countedPlay}; extra time, overtime, shoot-outs, and tiebreaks count only if expressly included. The Final Value is {participantA}'s {measure} plus {participantB}'s {measure}, each taken over Counted Play only. It and {line} use {measure}'s units and scale. Reported values use published precision. Derived measures are calculated from underlying statistics without rounding. Unclear units or calculations leave the Final Value unestablished. Completion requires all Counted Play to conclude. For full-contest scopes, normal endings under rules in force at scheduled start, including mercy rules and unneeded periods, count. Abandonment, retirement, weather-shortening, forfeits, and awarded results do not complete unfinished Counted Play. Postponement, suspension, rescheduling, or replay after incomplete Counted Play retains the participants and deadline; lineups may change. Resumed play continues the record; a full replay replaces it. An individual participates only by entering Counted Play. The market resolves to 0.5 if completion, the Final Value, or any individual participant's participation is unestablished at fixing. Otherwise, it resolves to Over if the Final Value exceeds {line}, to Under if below, and to 0.5 if equal. Only evidence public by {resolutionDeadline} establishes trigger time and source contents at fixing; resolution occurs only after that deadline. The trigger is the earliest verifiable official publication time for a record establishing completion, final statistics, and whether each relevant individual participated; absent that, its earliest verifiable capture time. Timestamps must prove contents existed then. Fixing is the earlier of one hour after the trigger and {resolutionDeadline}, or that deadline without a trigger. {officialSource}'s record establishes the facts. A verifiably newer version supersedes older versions of that record. Across official channels, an explicit correction or designation of authority prevails; other conflicts leave facts unestablished. Only versions public by fixing determine facts. Only if fixing is on {resolutionDeadline} may at least two independent reputable news organizations' original reports establish missing facts, with no unresolved material contradiction in reputable reporting. Syndicated copies count as one. Reporting cannot override official facts or resolve official conflicts. Changes after fixing and evidence after the deadline do not affect resolution. This market is not sponsored by, endorsed by, or affiliated with any entity referenced. Names and trademarks are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to a combined total in a single scheduled sports contest.
- countedPlay: string
- line: uDecimal
- measure: string
- officialSource: string
- participantA: string
- participantB: string
- resolutionDeadline: dateTime
- scheduledStart: dateTime
- sport: string
Sides: Over / Under
sportsTournamentNoWinner2
No winner
This outcome resolves to Yes if the parent question's fallback conditions apply, and otherwise resolves to No.
sportsTournamentParticipant
{participant}
This outcome resolves to Yes if {participant} is named the sole winner of the Tournament under the parent question's rules, and resolves to No once the parent question's rules show that {participant} can no longer win the Tournament or that the fallback conditions apply.
- participant: string
sportsTournamentWinner
{season} {competition} winner
This {sport} market has one named outcome per listed participant in the {season} {competition} (the "Tournament") plus the fallback. Exactly one outcome resolves to Yes; every other outcome resolves to No. The Tournament Result is the final result of the Tournament, including every contest, period, and tiebreak procedure that {officialSource} treats as part of the Tournament, as determined under the source rules below. A named outcome resolves to Yes if the Tournament Result names the corresponding participant the sole winner. A named outcome resolves to No once {officialSource}'s record shows that the corresponding participant can no longer win the Tournament under its rules, including upon elimination from the Tournament. A named outcome that has resolved to No is final and is not affected by any later change in {officialSource}'s record, including reinstatement of the corresponding participant. If any part of the Tournament is postponed, delayed, suspended, or rescheduled, the Tournament as ultimately completed applies if completed by {resolutionDeadline} UTC. If {officialSource} names a listed participant whose outcome has not already resolved to No the sole winner without a completed Tournament, including but not limited to a decision following abandonment, walkover, forfeit, disqualification, or withdrawal, that participant's outcome resolves to Yes. The fallback resolves to Yes and every named outcome not already resolved resolves to No if the Tournament Result records no winner, names co-winners, names anyone other than a listed participant the winner, names a participant whose outcome has already resolved to No, or is not established by {resolutionDeadline} UTC. The fallback resolves to Yes via a named outcome called No winner that is registered after every other named outcome has resolved to No. A result or status is established when {officialSource} publishes it or, failing that by {resolutionDeadline} UTC, when a clear consensus of independent reputable reporting available by that deadline supports it; otherwise it is not established. The Tournament Result is fixed as {officialSource}'s record stands one hour after {officialSource} first publishes on any official channel the final result of the Tournament or a winner by decision, or at {resolutionDeadline} UTC if earlier or if it rests on reporting; later appeals, corrections, reversals, or reassignments do not change it. This market is not sponsored by, endorsed by, or affiliated with {officialSource} or any competition, league, governing body, organizer, team, club, participant, or other person or entity referenced in this market. All names and trademarks are the property of the respective rights holders and are descriptive only. metadata=category:sports|subCategory:{sport}|semanticRestriction:This market must refer to the overall winner of a sports tournament.
- competition: string
- officialSource: string
- resolutionDeadline: dateTime
- season: string
- sport: string
How to trade a HIP-4 market
- 1
Fund a wallet on Hyperliquid
HIP-4 markets are quoted in USDC on HyperCore. Bridge USDC across from any supported chain, or deposit natively. Everything executes through your own wallet; there is no platform account.
- 2
Pick a market
Choose between single-strike price questions, multi-strike price ranges, and custom event markets with a written resolution clause. Each market shows its current odds as a price between $0.00 and $1.00.
- 3
Read who settles it, and when
Every market states its settlement authority and its expiry in its own on-chain description: protocol auto-settle from a price observation, a Hyperliquid validator vote, or a third-party deployer who signs the result. Check that tier before you size a position.
- 4
Place the order from your own wallet
Outcome tokens trade on a live spot order book, so you can use limit orders and exit before expiry at whatever the book will pay. Your position shares one account and one collateral balance with Hyperliquid perpetual futures.
Where the numbers live
Everything above is how the protocol works. For what it has actually done, including all-time and rolling-window volume, markets deployed and settled, the markets open right now with their current odds, and the live permissionless-deployment status, see the stats page. It is rebuilt from live exchange data and our own chain ingest every few minutes, and it publishes its methodology.
HIP-4 by the numbers
Volume, markets, settlements and permissionless status.
Frequently asked questions
What is HIP-4 on Hyperliquid?
HIP-4 is Hyperliquid’s native prediction market protocol. It creates outcome tokens that trade as hash-prefixed coins on the HyperCore spot order book, alongside every other spot asset, sharing the same account and the same collateral as perpetual futures. A market is a question with a defined settlement moment; each side of the question is a tradable token whose price sits between $0.00 and $1.00 and reads as the market-implied probability.
How do HIP-4 prediction markets settle?
A binary result pays one unit of the quote token to the winning side and zero to the losing side. Standalone outcomes can also have fractional payouts. Expiry and the recorded settlement action are distinct: a deployer-settled market needs its authorized settlement action. Before trading closes, exiting depends on available order-book liquidity.
Who decides the outcome of a HIP-4 market?
Check the individual market. Protocol markets can use a price observation or a validator vote. Permissionless markets, including price and event templates, are settled by their deployer or an authorized sub-deployer. The question being about a price does not by itself mean settlement is automatic.
Can anyone deploy a HIP-4 market?
Permissionless HIP-4 deployment is live on Hyperliquid mainnet. Anyone who meets Hyperliquid’s current staking and registration requirements can register as a deployer and launch markets from validator-approved templates. The deployer signs the final result for every market they create, so traders should check the settlement authority before trading. Liquidiction reads the deployer and template registries live and lists new markets automatically as they launch.
Is there a dispute period if a HIP-4 market settles wrongly?
The documented deployer settlement flow does not include an escrowed challenge window. The deployer can authorize a sub-deployer to submit settlement actions for its venue; this is not an open dispute process. Deactivation requires no active outcomes and completion of the minimum staking duration. Read the market’s terms and settlement authority before trading.
What is a HIP-4 template?
A template is a validator-approved blueprint for the display wording, side names and typed inputs of a market. The deployer supplies its values. Templates constrain market structure; they do not guarantee a correct settlement. The live registry below shows the available templates.
What is the difference between HIP-3 and HIP-4?
They are different protocols that are easy to confuse. HIP-3 lets third parties deploy their own perpetual futures markets on Hyperliquid, which is how the network lists perps on equities, commodities and indices. HIP-4 is the prediction market protocol: outcome tokens on the spot order book that settle to $1 or $0 at a defined moment. They connect at one point, which is that a HIP-3 perp can be the underlying a HIP-4 price market references.
Can I trade HIP-4 markets with leverage or hedge them?
HIP-4 outcome tokens are spot assets, and they share one Hyperliquid account and one collateral balance with perpetual futures. That shared collateral is what lets you run a perp against an outcome position rather than funding a second venue to hedge, and it is the main structural reason to trade prediction markets on a derivatives exchange.
Primary sources
These Hyperliquid references describe the protocol mechanics. For a particular market, also read its deployed terms and recorded settlement.
- HIP-4 outcome mechanics — outcome tokens, fractional payouts, merged Yes/No liquidity, and questions.
- HIP-4 deployer actions — validator-approved templates, settlement actions and sub-deployer permissions.
- Order behavior — IOC orders cancel any unfilled remainder; GTC orders can remain on the book.
Related: HIP-4 stats, deployers, all prediction markets, explore, bridge to Hyperliquid, perpetual futures, hedging, and how this compares to Polymarket.
