On this page (7 sections)
A liquidation heatmap is a price chart with horizontal bands of colour laid over it. Each band marks a price where, by the map’s estimate, leveraged futures positions would be force-closed, and the brighter the band, the more position value the model puts there. Bands under the current price come from longs, which the exchange liquidates by selling, and bands above it come from shorts, liquidated by buying.
What a liquidation map estimates
Exchanges don’t publish the liquidation price of each open position, so every map is built from public data and a model. The usual input is open interest, the total size of futures positions still open. When open interest rises while price trades at a level, the model counts new positions opened there, a long and a short for each new contract. It then assumes a mix of leverage for them, from around 10x up to 100x, and places each slice at the liquidation price that leverage implies. Providers weight that mix differently.
On isolated margin, the price for a long is roughly entry × (1 − 1/leverage) ÷ (1 − MMR), where MMR, the maintenance margin rate, is the small slice of position value the exchange makes you keep. At 25x and a 0.5 percent rate that is about 3.5 percent under the entry, as the liquidation price calculator shows, and a short mirrors it above.
Nothing rests at a band’s price the way a limit order rests in the order book. The band turns into market orders only if price reaches it while those positions are still open on the same margin.
Why price often reaches big clusters, and when it doesn’t
When price reaches a dense band, the liquidations fire as market orders in the direction of the move: long liquidations sell into a drop, short liquidations buy into a rally. That pushes price further and sometimes into the next band, which is how a liquidation cascade runs, and why traders call big clusters magnets.
Many clusters are never reached. There is usually a band on each side of price, and the map doesn’t say which one comes first. Traders close positions or add margin before liquidation, and the model sees none of it, so a bright band can be mostly empty by the time price arrives. Trend still sets the direction, and in a steady 4h uptrend a band under price can sit untouched for weeks.
Reading the map with structure
A band matters most when it sits just beyond a level the chart already makes obvious, such as the last confirmed swing high or low, or the previous day’s high or low. A swing counts as confirmed once five candles have printed on each side of it. Stops already sit past those levels, so a move through one fills them and then the forced orders behind, which is the fuel for a liquidity sweep: a move through the level that closes back inside. A band in open space, far from any swing, gives you no level to judge the reaction against, so we leave it out of the plan.
Whether there is a trade still comes down to the close on your timeframe, as it would without the map. With no close beyond the level the move was a sweep, and the trend flips only on a close beyond the last confirmed swing, however much of a band a wick took out on the way. Keep your own stop out of a band too, since a stop inside one fills while the band is firing, often with slippage.
A worked example on the 1h
The numbers are invented. ETH on the 1h is in an uptrend. A leg from a higher low at 3,120 closed above the 3,190 swing high, a break of structure, and topped at 3,260, now a confirmed swing high. The drop from 3,260 has come in one leg with no new swing low, so 3,120 is still the last confirmed swing low and the 1h trend stays up until a candle closes below it.
The map shows two bright bands. The lower one, 3,085 to 3,100, sits just under the swing low and is where the model puts 25x longs opened on the breakout: from entries between about 3,198 and 3,213 at a 0.5 percent rate, liquidation lands in that range. The upper one, 3,270 to 3,285, sits just over the high and comes from shorts opened on the way down.
A 1h candle drops to 3,090, through the swing low and into the lower band, and closes at 3,132, back above 3,120. The wick ran through the stops and the top of the band, and no candle has closed below 3,120, so this is a sweep. The next hour dips to 3,124 and closes at 3,140, and we buy that close.
- Entry: 3,140.
- Failure exit: a 1h close below 3,120.
- Stop: 3,070, 20 under the sweep low and past the bottom of the band. 70 points, 1R.
- Size: 0.684 ETH, so a stop-out costs 50 USDT including fees of 0.05 percent per side.
- Leverage: 18x at most, which puts liquidation near 2,980.
- Target: 3,245, just under the 3,260 high. 105 points, 1.5R.
The band’s bottom slice, 3,085 to 3,090, may still be open, so a stop a few points under the wick, at 3,087, would sit inside it; 3,070 is 15 points past it. The stop is about 2.2 percent from entry, so liquidation would reach it at about 36.8x; halving that and rounding down, as the leverage guide does, gives 18x and an estimate about 90 points past the stop. The upper band would add forced buying if price got there, and nothing in the map says it will, so the target goes a little under the 3,260 high.
Choosing the lookback
Most maps let you set how far back the model counts new positions, from a day to a few months. A short lookback shows only positions opened recently, which suits a 15m or 1h trade. A longer one adds older clusters that matter on the 4h and daily, though more of those positions have closed since without the model knowing. The colour scale is usually relative to the biggest cluster in view, so one band can look bright on a 24-hour lookback and faint on a three-month one. Compare bands within one lookback, and fit it to the trade: for the 1h example, a few days to a month.
Where the map is wrong
- Leverage is assumed. Open interest never says at what leverage a position was opened, so every band’s price rests on the model’s mix.
- Coverage. A map only sees the exchanges whose data it pulls, so positions elsewhere, decentralised exchanges included, are missing.
- Cross margin. The whole account balance backs a cross-margin position, so its real liquidation sits further out than the isolated formula puts it and moves as the balance changes.
- Mark price and tiers. Exchanges liquidate on the mark price, an index-based price that can lag a wick on one exchange, and large positions sit on higher MMR tiers that bring liquidation closer to entry.
Finora’s /liqs
Finora’s /liqs estimates liquidation levels for Bitcoin and other coins, modelled from 30 days of hourly data, and /liqs btc gives the nearest levels on one coin. Send a symbol and timeframe, for example ETH 1h, and Finora fetches the chart and replies with an entry zone, the targets in order and the invalidation level to set those estimates against.
Related reading
- How to tell a liquidity sweep from a real breakoutA liquidity sweep fills the stops behind an obvious high or low, then closes back inside. How to tell it from a breakout, with a 1h trade worked through.
- Liquidation price calculatorWhere an isolated long or short gets liquidated at your leverage.
- Crypto funding rates todayCurrent funding rates for liquid Binance USDT-M perpetuals: the highest and lowest rates, each contract’s payment interval and the next funding time.
- How much leverage to use, worked out from your stop-lossWhen size comes from the stop, leverage only sets margin and liquidation. How much leverage a stop distance allows, with a worked 4h BTC example.