Learn · guide

How to read a liquidation heatmap (and what it can't tell you)

A liquidation heatmap shows estimated price levels where leveraged futures positions would be forced to close. Bright bands above price are mostly shorts that would be liquidated; bright bands below are longs. It is a model, not real orders - and in our own test, price did not move toward the biggest clusters more often than chance.

Live example, BTC (22 Oct 2026, 01:40 UTC): price 83,807. The biggest modeled cluster above is at 85,036 (+1.47%, about $119.6M of shorts). Below: 79,369 (-5.30%, about $83.0M of longs). Within 5% of price the model holds $462.2M of longs below and $811.3M of shorts above. Open the live map.

What a liquidation heatmap shows

Every leveraged futures position has a liquidation price: the level where losses eat the margin down to the exchange's maintenance requirement and the exchange closes the position by force. A long is closed with a market sell, a short with a market buy.

Exchanges do not publish where all these liquidation prices sit. A liquidation heatmap estimates them: time runs left to right, price bottom to top, and brightness shows how much position value would be liquidated at each price level at each moment. When price trades through a level, those positions are gone - liquidated or closed - and the band ends.

How the map is built

Most heatmaps follow the same recipe. Ours is published in full, so you can see every assumption:

1. New positions. When open interest rises, new positions were opened. We place them at the average price of that 5-minute bar.
2. Longs vs shorts. Every contract has a buyer and a seller, so open interest alone cannot say who opened. We split new positions by that bar's aggressive buying vs selling (taker volume).
3. Leverage. Nobody knows the real mix. We assume 10x, 25x, 50x and 100x in shares of 35%, 30%, 20% and 15%, with 0.4% maintenance margin.
4. Liquidation price. Long: entry × (1 − 1/leverage + 0.4%). Short: entry × (1 + 1/leverage − 0.4%).
5. Closing. When open interest falls, all estimated positions shrink proportionally. When price trades through a level, that level is removed.

Change any assumption - the leverage mix, the long/short split, the exchanges included - and the bright bands move. That is the main reason heatmaps from different sites disagree, and why none of them should be read as a list of real orders.

Reading it, step by step

1. Find the current price. On our map it is the white line with a yellow marker on the right axis.
2. Look above and below separately. Above price: mostly shorts that would be forced to buy. Below: mostly longs that would be forced to sell.
3. Compare sizes, not colors. Colors are relative to the rest of the chart. The key-levels table under our map gives the estimated USD size and distance of the five biggest clusters on each side.
4. Check the balance near price. "Within 5%: X of longs below vs Y of shorts above" tells you which side has more estimated leverage close by.
5. Mind the time window. A 14-day view shows old, large bands that may already be closed in reality; a 3-day view shows what built up recently.
6. Watch bands end. A band that stops at a candle means price traded through it - the model treats those positions as liquidated. Real liquidations for the same hour are on live liquidations.

Liquidation map, liquidation heatmap, liquidity heatmap

The names are used loosely. A liquidation map is usually a snapshot: estimated liquidation volume per price level right now. A liquidation heatmap is the same estimate over time. A liquidity heatmap, strictly, shows real resting limit orders in the order book - see our liquidity heatmap - although some sites use the term for liquidation maps too. Only the order book version is real data; the other two are models.

Does price really go to the clusters?

The most repeated claim is that price is "pulled" toward big liquidation clusters because large players hunt them. We wrote down a test before looking at the data: every 4 hours, take the biggest cluster 0.5-3% from price on each side; if one side is at least 1.5x bigger, does price touch it before a level at the same distance on the other side?

55.2% sits inside what random side assignment produces (up to 61% in 99% of shuffles). The split by direction gave it away: "cluster below" won on both coins, "cluster above" did not - the market drifted down that month. A forward test on new data is running. Full test.

Sensible ways to use it

• As a map of leverage, not a forecast. It shows where a move could turn into forced buying or selling, which can make a move faster if price gets there.
• For your own risk. If your liquidation price sits inside a big band, many traders share it; check the arithmetic in the liquidation price calculator.
• Together with real data. Rising open interest means new leverage is entering; extreme funding rates show which side is crowded; live liquidations show what was actually forced out.

Common mistakes

• Treating bands as orders. They are estimates; nobody can see other traders' liquidation prices.
• Trusting exact sizes. "$131M at 84,527" is a model output that depends on assumed leverage. Compare sides and distances, not decimals.
• Assuming one exchange is the market. Our model uses Binance open interest; positions elsewhere sit at other levels.
• Reading the magnet story as fact. It has not survived our test so far.

FAQ

Live maps: BTC liquidation heatmap · ETH liquidation heatmap · SOL liquidation heatmap · XRP liquidation heatmap · DOGE liquidation heatmap · all 20 coins