A bright band on a liquidation heatmap can look like a large order waiting to execute. Often it is a model estimate, built from incomplete information about where leveraged positions might become vulnerable. The first question is what the color actually measures.
What the colors measure
Some services infer liquidation zones from trading activity and assumed leverage. Others can observe public accounts more directly. Even then, the estimate depends on current collateral, other positions and the market’s liquidation rules.
There is a particularly important unit distinction in CoinGlass’s Model 2 documentation: its intensity values are relative model scores, not dollars of position notional, margin or expected liquidation. A displayed “100M” should not be read as $100 million waiting to be liquidated. Adding nearby bands does not produce a reliable liquidation-dollar total.
That does not make the map useless. It changes the question it can answer. A relative score may help locate areas the model considers unusually concentrated; it cannot automatically tell you how many dollars of compulsory orders will reach your venue.
From a vulnerable position to an actual sale
A leveraged long becomes vulnerable when its equity no longer meets maintenance requirements. The exchange then attempts to reduce or transfer risk according to its rules. On Hyperliquid, the trigger uses mark-based account values, and liquidation can proceed through the order book before a backstop takes over. Its liquidation documentation also explains why cross-margin positions affect each other.
A trader can add collateral, reduce the position or hedge before the estimated threshold is reached. Funding and other positions can change account equity. A map saved an hour ago therefore describes a possible future under assumptions that may already have changed.
The displayed last price also need not be the liquidation trigger. A brief trade through a colored level may leave the relevant mark price above it. Conversely, account deterioration elsewhere can bring liquidation closer without a dramatic move in the chart you are watching.
A cascade needs both flow and weak absorption
Imagine price approaching an estimated cluster of vulnerable longs near $96. If forced selling begins, its effect depends on who is willing to buy. If buyers replenish quickly, considerable selling may occur with little further decline. If bids retreat, modest selling can move price sharply and threaten another set of accounts.
The second case can become a cascade: falling prices weaken accounts, forced reductions add selling, and that selling pushes prices lower. The first case is absorption. The map alone cannot tell you which response will occur.
This is why a cluster is not a mechanical price magnet. It describes potential flow conditional on price and account state reaching a threshold. It does not supply the initial reason for price to travel there, nor guarantee that traders can profit from following it.
Comparing the estimate with the move
Save the map before the move, including its time window, venue coverage and legend. Write down the proposed mechanism: for example, selling may accelerate if the $96 area is reached while bids are already thinning. Then observe actual liquidation reports, open-interest changes and price response on a matching time window.
If substantial selling arrives but price stops falling, that is different information from a rapid decline through several levels. If nothing happens, the estimate may have been stale, misinterpreted or simply wrong. Preserve those misses alongside the impressive-looking hits.
The $96 band matters only if vulnerable positions remain open, their liquidation conditions are reached, and the resulting sales exceed the bids available. Without those links, the bright band is an estimate on a chart, not an explanation for the next price move.