GMX vs Hyperliquid for Trading Competitions

Two wallet-only perp venues with very different fee models and market structures. Comparing tracking, market groups, scoring cadence and format fit.

By VoltradePublished August 31, 20266 min read

GMX and Hyperliquid share the property that matters most for competition entry: neither requires you to link anything. On both, the wallet you register with is the address that gets scored. There is no OAuth, no read-only API key, no UID to find and paste.

After that they diverge on almost everything — market structure, fee model, what kind of competition each hosts well, and how the scoring job reads them.

Eligibility: both wallet-only, for different reasons

On Hyperliquid, the trading address is your EVM wallet, and the order book is fully on-chain. Scoring reads your fills and open positions straight from the public /info endpoint. No credential of any kind appears in the path.

On GMX, the same conclusion arrives by a different route: the registered wallet is the GMX trading address, so the sync reads volume per CampaignParticipation.walletAddress against the GMX API with no linking step. The venues directory lists GMX on Arbitrum and Avalanche, and the sync queries by source chain ID — defaulting to Arbitrum's 42161.

For a sponsor this means both venues have the lowest possible entry friction: connect wallet, register, trade. That is worth real turnout, and it removes the failure mode where a registrant silently scores zero because they never completed a linking step.

Side by side

GMXHyperliquid
EligibilityWallet-only — registered wallet is the GMX addressWallet-only — HL address is your EVM wallet
Data sourceGMX API, queried per address with a source chain IDPublic /info fills and open positions
Sync cadenceEvery 20 minutes (10,30,50)Every 5 minutes
Day bucketsStart-anchored 24h windowsStart-anchored 24h windows
Settlement window after end30 hours6 hours
Distinctive scoring featureMarket groups with per-group minimums and a qualifier capFills plus live position marks, for PnL
Formats run on VoltradeVolume competitionsPnL challenges
MarketsPerpetual and spotPerpetual
ChainsArbitrum, AvalancheHyperliquid L1
Directory fee entry0.01% maker / 0.06% taker, annotated as dynamic by market, pool and action0.02% maker / 0.05% taker, annotated as a volume-tiered base tier

Two rows carry most of the practical weight.

The cadence gap is a consequence of format. Hyperliquid's job runs four times as often because it scores PnL, and a PnL leaderboard moves with the mark price rather than only with new fills — an untouched open position changes value continuously. A volume board only moves when someone trades, so GMX at 20 minutes has ample resolution.

The settlement windows invert for the same reason. Hyperliquid scores through a 6-hour buffer after the end, with the final pass clamping the window to the end time so late fills cannot rewrite a settled result. GMX keeps discovering recently-ended competitions for 30 hours, because each day period stays open for 3 hours past its own end and a competition ending near the start of its final period needs a safe outer bound to finish closing.

GMX's distinguishing feature: market groups

The GMX sync supports something no other venue integration does in the same shape — market groups with per-group minimums and a qualifier cap.

That matters for a specific competition design problem. A plain volume competition rewards raw turnover, which concentrates activity wherever it is cheapest to generate. If you want a campaign to spread across several markets rather than pile into one, you need scoring that requires participation in more than one place. Market groups express that directly: define the groups, set a minimum per group, cap how much any one qualifier contributes.

If your campaign brief is "get real activity across our top five markets, not a leaderboard of one pair", GMX is the venue whose integration was built for it.

Hyperliquid's distinguishing feature: PnL that reconciles

Hyperliquid is the platform's cleanest venue for profit competitions, and the reason is data availability rather than any preference.

Challenge scoring needs three things per participant: realized PnL, unrealized PnL on open positions, and cumulative volume. One unauthenticated endpoint provides all three. The engine then applies a flat-at-start gate, cumulative volume brackets, max-cap pro-rata on realized PnL, an unrealized snapshot pro-rata and a minimum-volume eligibility check, and ranks by absolute total PnL. Participants who held positions at the start and never went flat, or who finished below the volume minimum, stay visible on the board but are excluded from payout at settlement.

The integration is honest about its one blind spot rather than hiding it. The fills endpoint returns at most 2,000 per call, and the job carries an explicit truncated flag — but it does not abort the tick. The partial window is scored and the flag is counted, which matters because pages ascend from the window start, so a truncated fetch is missing its latest days and reads exactly like "this trader stopped trading". The flag earns its keep at settlement instead: a final pass carrying any truncated or errored participant is deferred and retried, and once the grace window expires it settles regardless, marked DEGRADED, so one permanently broken wallet cannot freeze everyone else's payout.

Fee models, and why the numbers differ in kind

This is the comparison where the catalog numbers mislead if you read them as like-for-like.

The directory lists GMX at 0.01% maker / 0.06% taker with an explicit annotation: fees are dynamic by market/pool and action (swap/open/close/borrow), not fixed maker/taker. GMX also charges borrow fees driven by pool utilisation. Hyperliquid is listed at 0.02% / 0.05% with its own annotation: base tier; actual fees are volume-tiered and can change.

So one venue's number is an approximation of a dynamic, action-dependent schedule, and the other's is the top of a volume ladder most serious competitors will not be paying. Comparing 0.01 to 0.02 as though they were the same kind of quantity is a mistake.

The practical guidance for a sponsor: on a volume competition, the participants who dominate the leaderboard are the ones with the lowest effective cost, and a headline tier describes none of them well. Check the venues directory for the current catalog entry and the venue itself for what you would actually pay.

Which to pick

Ask what you are scoring first, because that decides it more than anything else.

  • Scoring profit? Hyperliquid. The data model fits the format, and it is one of only two venues on which Voltrade runs PnL challenges — the other is Lighter, compared in Hyperliquid vs Lighter.
  • Scoring turnover across several markets, with structure? GMX. Market groups with minimums and a qualifier cap are the reason.
  • Want spot as well as perps? GMX is listed for both; Hyperliquid is perpetuals only.
  • Want the fastest-moving public leaderboard? Hyperliquid, at a 5-minute cadence.
  • Optimising purely for turnout? Either. Both are wallet-only, which is the single biggest lever on registration-to-participation conversion.

Volume competitions on both venues are admin-launched: permissionless creators — humans in the launch wizard and autonomous agents on the API alike — are restricted to on-chain pool tracking, so a venue volume competition is arranged through Voltrade. PnL challenges split the two. Hyperliquid is one of the two venues in the supported challenge set, and challenge creation bypasses the restricted-creator gate, so anyone can launch a Hyperliquid challenge; GMX is not in that set, so GMX stays admin-only in both formats. Start at the launch page.

The single-venue write-ups go further: GMX competitions and Hyperliquid competitions. For the format decision underneath the venue decision, read volume vs PnL competitions, and for the shared mechanics, how DEX trading competitions work.

venue-pagecomparison

Keep reading

Every trade is a competition

Join a live volume competition or PnL challenge across top venues — or launch your own in minutes.