Ethereum (ETH) liquidation heatmap

How to read the liquidation map

A liquidation is the forced closing of a leveraged position when the trader runs out of collateral. The exchange does not ask for consent: it sells the long or buys back the short at market. One liquidation means nothing, but they gather into clusters, because thousands of people enter at similar levels with similar leverage. The map shows where those clusters sit.

Why price gravitates towards them. Liquidating a long is a market sell, which pushes price lower and triggers the next liquidation. The result is a cascade: one move feeds itself until the positions run out. That is why a large cluster acts as a magnet — not because of a conspiracy, but because forced liquidity is concentrated there for someone to take.

Leverage and distance. The higher the leverage, the closer the liquidation level sits to the current price: 100x positions close on a move of about one percent, 5x positions on a move of twenty. The leverage range on the map is adjustable: a narrow high-leverage range shows nearby clusters, a wide one shows the market-wide picture.

This is a model, not a fact — and that matters. Exchanges do not publish where participants' positions sit; nobody provides that data. We estimate the levels from accumulated volume and a leverage range. A computed level coinciding with a real cascade is probabilistic. Any service promising you “traders' exact positions” is either computing the same kind of model or selling you a confidence it does not have.

How to use it in practice. A cluster is not a target and not a forecast; it is a zone where a move can accelerate. It is most useful together with the order book heatmap: if there are no large limit orders below price and a dense liquidation cluster at the same time, a fall in that direction meets little resistance. If the cluster is shielded by a large wall, the path to it costs more than it looks.

Three mistakes almost everyone makes. The first is treating a cluster as a promise — “price will definitely go there”. It will not necessarily, and it owes you nothing. A cluster only says that if price does arrive, the move may get sharper. The second is reading the map without regard to the leverage range: nearby 100x clusters live for minutes, distant 5x ones may not trigger for months. The third is confusing liquidations with ordinary stop-losses. A stop is set by the trader and can be removed; a liquidation is executed by the exchange and asks nobody.

Why clusters build up below price in a rising market. When the market has been rising for a while, most positions are leveraged longs, and their liquidation levels sit below the current price. Hence the familiar pattern: a slow grind up and a sharp drop that looks “unprovoked”. Usually the drop simply passed through a dense zone of forced selling. In a falling market the picture mirrors: density gathers above, and the market travels in sharp squeezes upward.

What to look at alongside the map. Realised liquidations over the past day are a fact, not a model: the exchange publishes them as a stream. Comparing actual liquidations with computed clusters shows how well the model fits this coin right now. If actual triggers land noticeably above or below the computed ones, the leverage range is worth adjusting.

Viewing the map alongside candles, the order book and indicators is easier in the terminal. The model's assumptions are listed on the methodology page.