See where market makers are positioned — and where they break.
The perpetuals analog of options GEX. We classify the entire Hyperliquid account universe, isolate the market makers — the perp "dealers" — and map where their inventory sits, where forced unwinds cluster (the gamma profile), and the gamma-flip level where dealer liquidation pressure crosses. Built from exact on-chain liquidation prices, refreshed every few minutes.
Net inventory, the gamma-flip level, and the full liquidation-density profile per coin.
Aggregate long minus short across every classified dealer in the coin. A heavy net-short book means makers are leaning against price — and are exposed if it runs.
The liquidation density of dealer inventory by price. Tall bars are clusters where forced unwinds stack up — the levels price gravitates to and accelerates through.
The price where net dealer liquidation pressure changes sign. Above it, unwinds amplify upside; below it, they amplify downside. The single level worth watching.
Perps have no literal options gamma. This is a behavioral analog built from the exact on-chain liquidation price of every classified dealer account ≥$100k: per coin, the gamma_profile (liquidation-density histogram, 0.5%-of-mark bins, ±50%), the gamma_flip (cumulative long-liq/short-liq crossover), dealer net inventory and the amplify / dampen / transitional regime flag.
The gamma profile is already a distribution — over price levels of forced flow. It answers "if price reaches level L, how much notional is force-unwound there?" — not "what's the probability price reaches L?". Options markets extract the second directly from strike prices (Breeden–Litzenberger); perps have no strikes, so there is no implied-distribution shortcut. Connecting the two is a model — and the exact-liq-price fuel map to build it from is data nobody else has.
The full-universe liquidation map (every account ≥$100k, not just dealers) is archived every 5 minutes; gamma_exposure snapshots since 2026-07-06 and the regime layer since 2026-06-26. None of it is backfillable — positioning is point-in-time — so the dataset is forward-only and grows every week. We say so rather than hide it.
Every /quant/gex coin carries a distribution_context block — and it's archived, point-in-time, in the gamma_exposure snapshots.
A raw density number means little on its own; "the 92nd percentile of the last 30 days" is actionable. Each coin's near-mark fuel density, distance-to-flip, normalized dealer skew, regime score and funding rate are ranked against their own trailing 30-day history. Percentiles stay null while a coin's history is warming (<2 days) — the history_days field tells you exactly how much exists.
Does amplify actually precede fatter-tailed moves? We measure forward return, realized-vol and tail distributions conditioned on the regime flag, pooled across the full perp universe, against a volatility-matched null — so the flag can't take credit for "high vol → big moves".
STUDY IN PROGRESS
The assessment publishes with measured lifts, confidence intervals and sample sizes — the same public-scorecard bar as the regime engine. Findings land here when the forward-only archive is deep enough to say something honest.
This page is a 3-coin teaser. Pro unlocks /quant/gex and /quant/positioning for the entire Hyperliquid universe — per-account trader tags (market_maker / whale / smart_money), net delta by type, and the full gamma profile + flip per coin. The tagged liquidation-map archive for backtesting is on Pro Plus.
curl "https://cryptodataapi.com/api/v1/quant/gex?symbol=BTC" \ -H "X-API-Key: cdk_live_your_key"