Guide

Modelled vs real liquidation heatmap: how each one is built

Most liquidation heatmaps model levels from open interest and assumed leverage. How a heatmap from real positions differs, and the limits of both.

Blockmetric ResearchUpdated 6 min read

On this page
  1. The short version
  2. How a modelled liquidation heatmap is built
  3. How a heatmap from real positions works
  4. What Blockmetric does
  5. Centralised exchanges: derived figures only
  6. Limits of both approaches
  7. How to read a heatmap from real positions
  8. Common questions

A modelled liquidation heatmap estimates where liquidations could sit: it takes open interest, price and a set of assumed leverage levels and turns them into hypothetical liquidation prices. A heatmap from real positions adds up the liquidation prices of positions that exist, which is only possible on venues that publish every position, such as Hyperliquid and Lighter. Both charts can look the same; only the second can be checked position by position.

The short version

Modelled heatmapHeatmap from real positions
InputOpen interest, volume and price of a venueOpen positions as the venue publishes them
Liquidation priceAssumed, one per leverage level the model picksReported by the venue or computed from its rules
Size per levelA share of open interest assigned to that levelThe summed size of the positions liquidating there
Wallet detailNone: there is no position behind a levelEvery level can be traced to the wallets in it
VenuesAny venue with open interestOnly venues with public positions
Main limitThe assumptionsCoverage of the tracked wallets

Live coverage: methodology page; measured numbers: this study.

How a modelled liquidation heatmap is built

Most liquidation heatmaps cover venues that do not publish positions, so they reconstruct levels from what the venue does publish: open interest, volume and price.

The usual method works in three steps:

  1. Find where positions were opened. When open interest rises at a price, the model treats the added contracts as positions opened there.
  2. Assume a leverage mix. The new open interest is split across a set of leverage levels, for example 10x, 25x and 50x, with weights the model chooses.
  3. Place a liquidation price per level. For each assumed leverage, a long is placed a fixed distance below the opening price and a short the same way above it. The sizes are added per price bucket and coloured.

Every level on such a chart is the output of those assumptions. Change the leverage mix or the weights and the bands move, though the market is the same. A modelled chart shows where liquidations could sit if the assumptions hold, not where they sit.

How a heatmap from real positions works

On some venues every position that is still open is public account state. Hyperliquid's info endpoint returns, for any address, each position with its size, entry price, position value, margin used, leverage type (cross or isolated) and its own liquidation price.1 Lighter publishes the same kind of record per account, including position size, average entry price, allocated margin, margin mode and liquidation price.2

With those records there is nothing to assume. A liquidation price is the mark price at which an account's equity falls below its maintenance margin and the venue starts closing the position.3 A heatmap from real positions reads that price for each position, adds up the USD size per price bucket and repeats this over time. The brightness of a cell is the size of existing positions that would be liquidated at that price at that time.

What Blockmetric does

Our heatmap is built from real positions. The liquidation price is the one the venue reports; for dYdX and GMX, which report none, we compute it from their margin rules and on-chain state:

  • Hyperliquid. We read the open positions of the wallets we track, wallet by wallet, and use the liquidation price Hyperliquid reports for each one. The largest wallets are read every few minutes, smaller ones less often; the schedule and the measured age per wallet size are on data freshness.
  • Lighter. Tracked Lighter accounts are read the same way, with the liquidation price Lighter reports. Lighter is its own layer on the coin heatmaps.
  • Snapshots. Every 5 minutes the tracked positions are added up into price buckets about half a percent wide. For the top 20 coins by open interest a finer ladder is stored every minute.
  • Daily snapshot of all positions. Shortly after 00:00 UTC we store one snapshot of every open Hyperliquid position, not only the tracked ones. It fills the 1D columns, which reach back as far as 15 December 2025.
  • Stored as measured. Past positions cannot be read afterwards, so nothing is filled in later. A time without a stored snapshot stays empty on the chart, and every value is measured rather than computed between snapshots.
  • Wallets in a band. Clicking a cell on the liquidation heatmap lists the tracked wallets whose liquidation price sits in that band now, each linked to its wallet page.

The full method is under liquidation heatmap method.

Centralised exchanges: derived figures only

Centralised exchanges publish open interest, prices and a stream of liquidation orders, but no positions and no liquidation prices. Their liquidation levels cannot be read, and we do not model them, so they are not part of the heatmap.

What we show for them is derived. An optional chart layer, "Open interest added at this price", values the change in their open interest every 5 minutes at the price where it happened. It shows where contracts were added, not their leverage or where they would be liquidated. Their liquidation totals appear on the liquidations page as their streams report them. How the venues differ is covered in perp DEX vs CEX liquidations.

Limits of both approaches

Heatmap from real positions:

  • Only venues with public positions. Exchange positions are missing, so the chart covers a part of the market, not all of it.
  • Wallets not yet discovered. Live columns cover the wallets we track. New wallets are found through the public leaderboard, large trades and the daily snapshot of every position, so a wallet can trade for a while before it is tracked. What we do not track is not on the map, which makes every level a measured lower bound.
  • Minutes old. A position can change between two reads of its wallet.
  • Cross margin moves the price. For a cross margin position the reported liquidation price depends on the whole account. Hyperliquid notes that it can still change through funding payments or changes in unrealised profit and loss of the account's other positions.3 Cross positions share collateral, while an isolated position uses only its own margin.4 The liquidation price we store is the one reported at the time of the read.
  • Open orders are not positions. Stop and limit orders count only once they fill.

Modelled heatmap:

  • Assumptions. The leverage mix, the weights and the opening prices are chosen by the model. Two models draw two different charts from the same market.
  • No wallet detail. A band cannot be traced to any position, so it cannot be checked.
  • No reaction to margin changes. Added margin, partial reductions or positions without real leverage leave no trace in the model.

How to read a heatmap from real positions

  • Bright bands are real size. A band is the summed USD of existing positions with their liquidation price in that range. Long levels sit below the price, short levels above it.
  • Check the age of a band. A band that has stayed bright for days is old positioning; one that appeared an hour ago is new. A band that fades before the price arrives means holders added margin or reduced size.
  • Look inside a band. The wallet list under a cell shows whether one wallet or many sit there. The walk-through with an example is in how to read a liquidation heatmap, and the map beside it is explained in liquidation map vs liquidation heatmap.
  • Positioning is not a forecast. A liquidation cluster shows where forced orders would appear, not whether the price gets there. Our scoreboard records in advance whether published clusters were reached.

The BTC heatmap, ETH heatmap and HYPE heatmap show the same data per coin. The Free plan shows every interval from 5m to 1D with up to 30 days of every coin; longer history and 1m columns are listed on pricing.

Common questions

What is a modelled liquidation heatmap? A chart that places hypothetical liquidation prices by splitting a venue's open interest across assumed leverage levels. Its levels are an estimate under those assumptions, not positions.

What does "real positions" mean on a liquidation heatmap? That every level is the sum of positions that exist, each with its reported or computed liquidation price. On Hyperliquid and Lighter those positions are public per wallet or account.

Why are centralised exchanges not on a heatmap from real positions? They publish no positions and no liquidation prices. We show their open interest changes and reported liquidations as separate, derived figures.

Is a heatmap from real positions complete? No. It covers the venues with public positions and, for live columns, the wallets we track, so every level is a lower bound. The daily snapshot covers every open Hyperliquid position once a day.

Can the liquidation price of a real position change without a trade? Yes. Added margin, funding payments and, for cross margin, gains or losses on the account's other positions all move it.

Sources

  1. Hyperliquid Docs, "Info endpoint: Perpetuals" (clearinghouseState) (accessed 2026-11-16)Back to text
  2. Lighter Docs, "WebSocket" (Position JSON, account_all_positions) (accessed 2026-11-16)Back to text
  3. Hyperliquid Docs, "Liquidations" (accessed 2026-11-16)Back to text
  4. Hyperliquid Docs, "Margining" (accessed 2026-11-16)Back to text

How we collect and check the data: Data & methodology.

See today’s numbersThe live pages update continuously.

Open the liquidation heatmap
LiveHyperliquid markets 12sOperationalHyperliquid positions & whales 18sOperationalLiquidation heatmap & levels 16sOperationalLighter markets & liquidations 20sOperationalLighter positions & heatmap 10sOperationalLighter RH positions & heatmap 18sOperationalBinance markets 1mOperationalBybit markets 14sOperationalBinance liquidations 19sOperationalBybit liquidations 19sOperationaldYdX positions & heatmap 13sOperationalGMX positions & heatmap 18sOperationalLong/short & open interest history 21sOperationalOrder books & trade flow 16sOperationalOn-chain transfers 12sOperationalSecurity monitor 12sOperationalBlockmetric Sentiment 10mOperationalAgency releases 29sOperationalAlert delivery 12sOperationalDataset files 1mOperational16:08:13 UTCData: Blockmetric Research