I got stopped out of a perfectly good long a while back, watched price reverse within minutes, and did the thing everyone does, which is complain about market makers hunting my stop. Then I pulled up a liquidation heatmap for the same pair and the complaint got harder to sustain. My stop was sitting in the middle of the brightest band on the entire chart. Nobody hunted me personally. I had parked my exit inside the largest pool of forced sellers on the map, and the market did what it does with pools like that, which is drink them.
Since then I do not set a stop on a perp trade without checking the heatmap first. It takes about ninety seconds and it has moved my stops more than any indicator I have used. Worth walking through what these maps actually are, why they work at all, and the exact sequence I run before a stop goes in.
What a liquidation heatmap actually shows
The first thing to get straight is that the map does not show real orders. Exchanges do not publish where everyone's liquidation price sits. The heatmap is an estimate built from two inputs, changes in open interest and assumptions about the leverage people typically use.
When open interest rises as price trades through a level, new positions were opened around there. The tool cannot see each trader's leverage, but the common tiers are predictable, mostly 10x, 25x, 50x and 100x, because those are the defaults on every slider. Apply each tier plus a maintenance margin estimate and you get the prices where those positions would be force-closed. A long opened at 25x gets liquidated roughly 4 percent below entry, a little less once maintenance margin is counted. Stack those estimates across every price where open interest built up, and the bright bands are the levels where lots of estimated liquidation prices pile up together.
So the map is a model, and it inherits the weaknesses of its assumptions. It cannot see cross margin accounts holding more collateral than the position suggests. It cannot see traders who add margin or close early, and levels decay as positions get closed, which the better tools age out imperfectly. I treat brightness as a rough probability that forced flow lives at a level, and I hold it loosely.
Why price gravitates toward the pools
A liquidation is a forced market order. When a cluster of long liquidations sits below price, that cluster is a pool of guaranteed selling that will hit the book the moment price touches it. For anyone who needs to buy size, that is attractive. Pushing price a little lower to trigger the pool means buying into a wave of forced sells instead of chasing offers higher, and the same mechanics run in reverse for short liquidations above price.
You do not need a conspiracy for any of this. Large traders are drawn to liquidity the way water finds a drain, and the densest liquidity on a perp chart is often the forced kind. Add the cascade effect, where the first liquidations push price into the next tier and trigger more, and you get the familiar pattern. Price grinds toward a bright band, accelerates through it in one violent candle, then very often reverses once the pool is drained, because the aggressive flow driving the move was eating exactly that liquidity and there is nothing behind it.
This is also why the classic stop placements fail so reliably. Round numbers, the most recent swing low, the bottom of the obvious range. Everyone can see them, everyone parks stops and liquidation prices near them, so those areas light up on the heatmap, so they become targets. The swing low that looks like protection on a bare candlestick chart frequently sits at the near edge of a pool, which is exactly where a sweep begins.
The sequence I run before any stop goes in
- Pull the heatmap on a window that matches your hold time. For a multi-day hold I look at a one-week map, since a 12 hour view hides older clusters that are still live.
- Mark where structure says the stop should go. Swing point, invalidation level, whatever your system uses. This part does not change.
- Check what sits between current price and that level, and just beyond it. The question is whether the stop lands inside a dense band or within sweep distance of one.
- If it does, move it past the far edge of the cluster, plus a buffer. I scale the buffer to volatility, something like a fraction of the average daily range, because sweeps tend to overshoot the band a little and the buffer exists for the overshoot.
- Re-run position sizing on the new stop distance. Wider stop, smaller size, same dollar risk. If the size gets too small to bother, skip the trade. Tucking the stop back inside the pool to keep the size up is how you convert someone's estimated liquidation map into your own realized loss.
- Check the other side too. On a long, a dense band of short liquidations above entry is a reasonable take-profit zone, since the forced buying there is likely to be your exit liquidity.
Using the same map for entries
The stop logic inverts nicely. If a bright band of long liquidations sits a couple of percent below price, I start treating that zone as a candidate entry rather than a hazard. The sweep into the pool delivers forced sellers, and buying from forced sellers is about the best fill available in a perp market. In practice that means resting bids just past the far edge of the cluster, or waiting for the wick through the band and the reclaim, then entering on the reclaim with a stop below the sweep low. The second version costs a worse price and buys confirmation, which is usually worth it once the pool is spent.
The main failure mode is treating the map as a magnet that always pulls, because it does not. In a strong trend the pools on the far side simply get abandoned and price never comes back for them. The heatmap tells you where forced flow is likely to sit if price gets there, and nothing about whether it will get there. I use it to veto bad stop locations and to pick entry zones, and I let other tools decide direction.
One more caveat from experience. Different tools build their maps with different leverage assumptions and decay models, so two heatmaps for the same pair can disagree. Learn the habits of whichever one you use before trusting it with stop placement. And each exchange has its own open interest and its own pools, so a map built on one venue says little about another, although the biggest clusters on the biggest venues tend to matter everywhere, since cascades spill across markets through arbitrage.
We built liquidation and open interest views into Blockcircle for exactly this pre-trade check, but the tooling matters less than the habit. Look at the map before the stop goes in, and assume the obvious level sits inside the sweep zone until the map shows otherwise. Over enough trades, the stops that survive are usually the ones placed one band further out than felt comfortable.