There is a number circulating this week that looks like a promise. It says that if Bitcoin trades up to $84,975, $1.273 billion worth of short positions will be forcibly closed. A second number says that if Bitcoin slips to $77,256, $848 million of longs will be wiped out. Together they form a clean map of where the leverage is stacked โ and, for the hopeful crowd, a clean map of where the money is too.
I have been reading these heatmaps for years, and the first thing I do now is not look at the color gradient. It is to ask what the number actually measures. The figure above $84,975 is not $1.273 billion of orders sitting in an exchange's matching engine. It is not money waiting in escrow. It is a model output โ an estimate โ and the gap between what it looks like and what it is happens to be the entire story.
Silence speaks louder than hype.
Context: how a heatmap is built, and what it isn't
Coinglass has become the industry's default source for liquidation data, and to its credit the methodology is public enough to be inspected. The company takes open interest from the major centralized exchanges โ Binance, OKX, Bybit, and others โ and multiplies it by an assumed distribution of leverage. From that multiplication it derives "liquidation intensity": a theoretical estimate of how much position value sits at a given price level, ready to be liquidated if price gets there.
The word to hold onto is "theoretical." No exchange publishes its real liquidation order book. They cannot, for competitive reasons, and they won't, for reasons that have as much to do with market-making as with secrecy. So the heatmap is an inference, a probability cloud dressed up as a wall. It tells you where leverage probably clusters. It does not tell you what will actually fire, or in what size.
This is not a knock on Coinglass specifically. It is the state of the art. When I spent six months in 2017 auditing time-crowdsale contracts in Warsaw, the lesson I kept relearning was that the difference between a number and its meaning is where people lose money. The code was verifiable; the marketing around it was not. Liquidation heatmaps live in that same gap.
The format itself is young. Heatmaps only became a mainstream trading narrative around 2023, as perpetual futures volumes outgrew spot and retail traders went looking for a visual shorthand for leverage. What emerged was not a new dataset so much as a new way of rendering an old one. The underlying open interest had always been available. The heatmap simply turned it into a picture, and pictures travel further than tables. That is the real innovation, and it is a marketing innovation as much as a technical one.
The leverage sandwich
Set the methodology aside for a moment and look at the structure, because the structure is genuinely informative.
At $84,975, the model clusters $1.273 billion of short-side liquidation intensity. At $77,256, it clusters $848 million of long-side intensity. The ratio is roughly 1.5 to 1 in favor of the shorts. The two levels sit about $7,719 apart โ roughly plus or minus 4.75% around a midpoint of $81,115 โ which is a band a liquid market can travel through in hours, not weeks.
That band is what traders call a liquidity sandwich. Price is wedged between two pools of forced flow. Move up, and the buying that liquidates shorts pushes price higher still, feeding itself. Move down, and the selling that liquidates longs does the same in reverse. Neither outcome is destiny, but the asymmetry matters: the upside pool is half again as large as the downside pool, which means the short squeeze, if it comes, has more fuel behind it than the long cascade has on the other side.
This is the part of the report that is worth taking seriously. Not the dollar figures โ the geometry. Shorts are more crowded than longs. Whatever the model's error bars, that relative reading is the one signal that survives scrutiny, because it is a comparison of two numbers built the same way. A ratio is more robust than either of its inputs, in the same way that a poll's margin is more reliable than its absolute percentage.
There is a second, quieter reading. If both levels are being emphasized at once, the person who published the snapshot was almost certainly looking at a price somewhere between them. You do not frame a sandwich unless you are standing inside it. That tells you the market was, at the moment of capture, undecided โ coiled in a narrow band, leveraged on both sides, waiting.
What the cascade actually looks like
It helps to be concrete about the mechanism, because the abstraction hides the risk.
A perpetual futures contract has no expiry. To keep it anchored to spot, exchanges charge a funding rate โ a periodic payment between longs and shorts. When funding is positive, longs pay shorts; when negative, shorts pay longs. The rate is the market's price for being on one side of the boat.
Liquidation happens when a position's margin falls below the maintenance requirement. At that point the exchange's engine closes it at market. If a cluster of shorts gets closed at once, those closes are market buys, and market buys move price up, which liquidates more shorts, which triggers more buys. That is a short squeeze. The long cascade is the mirror: forced market sells, lower price, more forced sells.
The asymmetry of the sandwich matters here because squeezes and cascades are not symmetric events in practice. A short squeeze tends to be violent and fast โ shorts are structurally more leveraged and more likely to use tight stops, because being short an asset with a long-term uptrend is psychologically expensive. A long cascade can be deeper but often unfolds in steps, as buyers step in at round numbers. If I had to bet on which side of this sandwich resolves with more drama, I would bet on the top.
There is also a layer most heatmap readers never see: the mark price. Liquidations are triggered not by the last traded price but by an index-weighted mark price, and each exchange weights its sources differently. That means a wick on one venue can liquidate positions that never saw the same wick on another. When people talk about "stop hunting," this is part of the machinery โ a sharp move on a thin book can cascade through mark-price feeds even if the broader market barely moved. Add auto-deleveraging to the picture, the mechanism exchanges use when the insurance fund cannot cover a shortfall, and you get a system where the visible price and the liquidation price are not always the same number.
But โ and this is the sentence I want you to underline โ "more drama" is not the same as "more money." The intensity number is a measure of potential, and potential is a polite word for a guess.
The contrarian read: the number that isn't a number
Here is where I part ways with most of the coverage.
The single most common error I see in crypto media is the transformation of "liquidation intensity" into "pending liquidations." A headline will say that $1.273 billion is "waiting to be liquidated" above $84,975. That is wrong in a way that changes how people trade. Intensity is a derived quantity: open interest times an assumed leverage distribution, summed across exchanges with different margin rules, different mark-price conventions, and different auto-deleveraging logic. It is a heat estimate. It is not a ledger.
The real liquidated amount, if price reaches $84,975, will almost certainly be lower โ often much lower โ than the intensity figure. Some of that notional will be hedged elsewhere. Some will be collateralized well enough to survive the touch. Some will belong to positions that get closed voluntarily before the level is hit, precisely because their owners read the same heatmap. And some will simply not fire, because price wicks through and back without sustained follow-through.
This is why I have grown wary of the phrase "liquidation magnet." It is half true. Dense liquidation zones do tend to attract price, because market makers and large players know where the forced flow is and can profit by pushing toward it. That is stop hunting, and it is as old as markets. But the same density that makes a zone attractive also makes it a place where sophisticated money sets traps. A wick into the zone that does not hold is a gift to whoever positioned against the crowd.
Truth is often buried under the noise. The noise here is the dollar sign. The truth is the word "intensity."
The missing context problem
I cannot in good conscience write about this data without flagging what is absent, because the absences are more consequential than the presences.
There is no timestamp. A leverage snapshot is perishable on the order of hours. Open interest can double or halve in a day; funding can flip sign in a session. A heatmap without a publication time is a map of a country that may have redrawn its borders since it was printed. If you are reading one that is more than a few hours old, you are reading history.
There is no funding rate. This is the biggest hole. Funding tells you which side is paying to stay in the trade, and therefore which side is crowded in a way that costs money. If funding is deeply negative โ shorts paying longs โ the short crowding is confirmed and the squeeze thesis strengthens. If funding is positive while short intensity dominates, something does not add up, and the heatmap is telling you less than it appears to. Without funding, the 1.5-to-1 ratio is a shape without a temperature.
There is no open interest absolute value. Without it, you cannot tell whether leverage is at a historical extreme or merely normal. A two-billion-dollar sandwich in a forty-billion-dollar OI market is a different animal from the same sandwich in a fifteen-billion-dollar market. The number means nothing without its denominator.
And there is no spot price, no volume, no volatility reading. Each of those would let you sanity-check the band. None is present. What you have is two coordinates and two intensities, which is enough to sketch a structure and not enough to trade it.
I learned this the hard way in 2020, when I was writing a risk-parameter guide for Aave and kept getting asked for a single "safe" number. There wasn't one. The safety of a position depended on collateral volatility, liquidation thresholds, and the behavior of the oracle โ three things that had to be read together. A number in isolation is a rumor with a decimal point.
Who actually benefits from the heatmap
Step back from the chart and look at the incentives, because they explain why these snapshots spread the way they do.
Coinglass sits in the data-middleware layer. It does not create trades; it shapes expectations about them. Its product is a public good with a marketing engine attached โ the daily heatmap is genuinely useful, and it is also genuinely shareable, which is why it travels. The propagation itself has an effect. When enough people read that $84,975 is the squeeze level, some of them position long ahead of it, which nudges price toward the level, which makes the level look prophetic. That is reflexivity, and heatmaps are unusually good at generating it.
The exchanges, meanwhile, are the structural winners of any liquidation event. Whether the sandwich resolves up or down, the clearing engines collect fees and the insurance funds absorb the residue. That is not a conspiracy; it is the business model. But it does mean that the interests of the exchange and the interests of the leveraged trader diverge precisely at the moment the trader most needs them aligned.
And the downstream effects run further than most retail traders realize. A violent squeeze feeds through to spot as forced buying. A cascade feeds through as forced selling, which can trigger on-chain liquidations in lending protocols where Bitcoin is posted as collateral, which can in turn stress synthetic stablecoins that hold delta-hedged positions. I watched a version of this in 2022, when I spent three weeks with a crisis team fact-checking rumors for a ten-thousand-member community during the Terra collapse. The lesson then was the same as it is now: the second-order liquidations are the ones that hurt, because nobody modeled them.
The takeaway
So what is this heatmap actually good for?
It is a risk map, not a direction signal. It tells you where the ground is thin. It does not tell you which way the crowd will run. The people who treat it as a prediction tend to be the ones who get stopped out on the wick.
If you are going to use it, use it the way a structural engineer uses a stress diagram. Note that shorts are more crowded than longs. Note that the band is narrow, which means the market is coiled. Note that the resolution, whenever it comes, will be faster than the setup suggests. Then go find the four numbers the heatmap left out โ funding, open interest, spot volume, and a timestamp โ because those are what turn a shape into a thesis.
Code does not lie, only humans do. The heatmap's code is honest about being an estimate. It is the humans reading it who quietly upgrade "intensity" into "certainty," and that upgrade is where the money goes.
Watch the funding rate first. If it turns deeply negative while price holds the middle of the band, the squeeze has a real engine behind it. If it stays positive while the shorts pile up above, the heatmap is showing you a crowd that is not paying to be there โ and crowds that are not paying tend to leave early.
The next few sessions will tell us whether the sandwich resolves with a squeeze or a fake-out. Either way, the number that mattered was never the two billion dollars on the chart. It was the word printed next to it, the one almost everyone skipped.