Guide

How to read an order book heatmap: walls, depth and pulled liquidity

How to read an order book heatmap: what the bright bands are, how walls are detected, what pulled and filled walls mean and where order book data stops.

Blockmetric ResearchUpdated 6 min read

On this page
  1. What an order book is
  2. Reading the heatmap
  3. How walls are detected
  4. Pulled, filled or out of view
  5. Depth, imbalance and the venues
  6. What an order book heatmap does not tell you

An order book heatmap shows where limit orders were waiting over time: each column is a moment, each row a price, and the brighter a cell, the more dollars of resting orders sat at that price. Bands that stay in place are liquidity that kept waiting; bands that vanish just before the price arrives were pulled. It is a picture of the visible book, not of the traders behind it.

This guide explains how to read the liquidity heatmap on a coin's order flow tab, how we detect walls and decide whether they were pulled or traded into, and what order book data cannot show. The taker side of the same tab, the CVD line, has its own guide: how to read CVD.

What an order book is

An order book is the list of limit orders that are waiting to trade. Below the current price sit the bids, orders to take the asset at that price or lower. Above it sit the asks, orders to hand it over at that price or higher. A trade happens when an incoming order crosses the spread and matches resting orders: the resting order is the maker, the incoming one the taker.

Exchanges publish the book in aggregated levels: the total size waiting at each price, without order ids or owners. Hyperliquid, for example, returns at most twenty levels per side in one book snapshot and can aggregate them to fewer significant figures for a wider view.1 Everything an order book heatmap shows is built from those totals.

Reading the heatmap

On the order flow tab, the liquidity heatmap puts the book of every stored moment next to the candles of the same window.

  • Rows are price buckets, columns are snapshots in time.
  • Brightness is the USD value of the orders waiting in a bucket. Below the price the bright cells are bids, above it asks.
  • Candles on top show what actually traded, so you can see where the price met a band and what happened there.
  • Boxes mark walls. A dashed red box is a wall that was pulled before the price got there.

A few patterns are worth telling apart:

  • A band that stays while the price approaches and then trades through it was resting liquidity that got used. The candles show the price passing the band, and the band fades as the orders are filled.
  • A band that stays while the price touches it and keeps standing is liquidity that is absorbing the flow: trades hit it and it does not give way. On the walls list this shows as "still there after the price touched it (absorbing)".
  • A band that disappears just before the price arrives was pulled. The orders were cancelled or moved before they could trade.
  • A band that moves with the price can be a set of orders being re-placed at a fixed distance from the price, so it never gets reached.

Depth thins out away from the price on almost every book, so a bucket far from the price looks dim even when it holds real size. That is why we compare a bucket with its neighbours, not with the busiest part of the book.

How walls are detected

A wall is a price bucket that holds far more resting orders than the buckets around it. Our definition, also shown on the order book walls page:

  • The bucket holds at least five times the median of its neighbouring buckets on the same side, measured against a smoothed baseline of recent checks.
  • It also has to clear a minimum size, higher for BTC and ETH than for other coins.
  • The innermost bucket next to the price never counts, because the top of the book is always the heaviest. At most three walls per side are tracked, the largest ones.
  • A candidate becomes a wall only after two checks in a row, five seconds apart. Orders that flicker in and out are dropped without a trace.

The walls list on the tab shows each wall with its venue, side, price range, size, distance from the price, the time it stood and what happened to it.

Pulled, filled or out of view

Every wall ends in one of three ways, and we decide which by looking at the trades between the checks:

  • Pulled before the price touched it. The wall disappeared and the price never traded into its bucket while it was there.
  • Price traded into it, then it was gone. The price reached the bucket and the wall shrank away. With aggregated book data we cannot tell whether the orders were filled by trades or cancelled at the moment of the touch.
  • Out of view. The bucket left the range we observe, because the price moved far away or the stream stopped.

A pulled wall is the pattern people often call spoofing: large orders shown to the market and cancelled before they can trade. The book cannot prove that. Orders are cancelled for ordinary reasons all the time: a market maker adjusting quotes after a price move, an algorithm reducing risk, a trader changing their mind. Without order-level data and the identity of the owner, a pulled wall is a size that was there and then was not, nothing more. What the record does show is how reliable the visible liquidity at a price was. Many pulled walls in one area mean the size you saw there often did not stay.

Depth, imbalance and the venues

Above the chart, the tab shows the dollars resting within one percent of the price on both sides, split into bids and asks, and whether more sits on the bid or the ask side. That is a quick read of how much size the price has to get through nearby. It changes from second to second, and a side that looks heavy can be pulled as quickly as it appeared.

The heatmap covers the books we stream:

  • Hyperliquid for the top coins by open interest, in two resolutions: a fine view close to the price and a coarse view further out.
  • Binance and Bybit for a few of the largest coins, from our own copy of their book kept from their live streams. Binance levels far from the price appear once they change; Bybit sends only the levels nearest to the price, so its view is narrower.

A venue switch above the chart shows one venue at a time or all of them together. A snapshot of the book is stored every 30 seconds, and walls are checked every five seconds.

On the Free plan, the tab shows the current book of BTC, ETH and HYPE with its walls. The heatmap over time, and every covered coin, come with Pro.

What an order book heatmap does not tell you

  • Who placed the orders. Order books carry no names or order ids. A wall cannot be attributed to a whale, an exchange or a market maker.
  • Intent. A large bid can be a real interest to trade, a hedge, a quote that will move with the price, or an order that will be cancelled in a second. The book looks the same.
  • Hidden size. Orders that are not displayed in the book, and trades on venues we do not stream, are not in the picture.
  • What happens next. Liquidity that sat at a price in the past is not a level the price has to respect. Walls get filled, pulled and replaced constantly.
  • Every change. We store a snapshot every 30 seconds, so orders that appear and disappear between two snapshots do not show in the heatmap, even though the five-second wall check may have seen them.

The most useful way to read the book is together with the two other views on the same tab: the taker flow that trades against the book, and the liquidation heatmap of the same coin, which shows where forced orders would come from. For the same order flow across all coins, the markets screener sorts every coin by its CVD.

Sources

  1. Hyperliquid Docs, "Info endpoint", section "L2 book snapshot" (accessed 2026-11-11)Back to text

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

See today’s numbersThe live pages update continuously.

See the BTC order book heatmap
LiveHyperliquid markets 20sOperationalHyperliquid positions & whales 18sOperationalLiquidation heatmap & levels 18sOperationalLighter markets & liquidations 25sOperationalLighter positions & heatmap 21sOperationalLighter RH positions & heatmap 16sOperationalBinance markets 35sOperationalBybit markets 1mOperationalBinance liquidations 24sOperationalBybit liquidations 24sOperationaldYdX positions & heatmap 25sOperationalGMX positions & heatmap 22sOperationalLong/short & open interest history 27sOperationalOrder books & trade flow 10sOperationalOn-chain transfers 18sOperationalSecurity monitor 1mOperationalBlockmetric Sentiment 2mOperationalAgency releases 2mOperationalAlert delivery 16sOperationalDataset files 23mOperational16:30:18 UTCData: Blockmetric Research