Liquidity grabs and stop hunts: what the liquidation tape actually shows

If you have watched a crypto chart for more than a week, you have seen it. Price spikes through an obvious level, exactly far enough to take out the stops sitting there, and then snaps back like the whole move was staged. Traders call it a liquidity grab, a liquidity sweep, or a stop hunt, and the usual explanation is some version of "the market makers ran the stops."

The explanation is half right, and the half that is wrong costs people money. Here is what the tape actually shows.

What a liquidity grab actually is

Start with the vocabulary, because the terms get used loosely and the differences matter.

Liquidity, in this context, means resting orders that someone else can trade against: stop-loss orders, liquidation triggers, and pending limit orders clustered at a price. They pool at predictable places. Just under an obvious support level sits a shelf of long stops. Just above a range high sits a shelf of short stops and the liquidation prices of everyone who sold the breakout early. Those pools are the "liquidity."

A liquidity sweep is price trading through one of those pools and filling the orders in it. Neutral term, no intent implied. A liquidity grab is the same event described by price-action traders who expect a reversal to follow. A stop hunt is the same event again, framed as deliberate: someone pushed price there on purpose to trigger the stops.

All three point at one mechanical fact: when price reaches a level where orders are stacked, those orders execute, and because stops and liquidations are market orders, they push price further in the same direction before it can settle. That is not a theory. It is how a matching engine works.

The part that is real: cascades on the tape

You do not have to infer the cascade. It prints.

A stop-loss becomes a market order the instant it triggers. A leveraged position that hits its liquidation price gets closed by the exchange's liquidation engine, also at market. Both dump into the book in the direction the move is already going, which is why a level that "should" hold sometimes tears through in seconds. The liquidation feed measures exactly this, per exchange, in dollars.

Here is the live tape at the moment of writing, straight from the cross-exchange liquidation feed:

Read those three rows together and the mechanism is right there. Over the day, BTC fell about 1.7% and the forced sellers were almost entirely longs: $64.5M of long liquidations against $9.9M of shorts. Then in the final hour price ticked up 0.98%, and the forced side flipped, 92% of the hour's liquidations were shorts getting run out. The liquidated side is not random and it is not personal. It follows price into whichever pool of leverage sits in the way.

Notice this happened with funding at the 18th percentile of BTC's 804-day history, a rate that rounds to nothing. Nobody was crowded by the usual funding tell. Longs still got flushed, because a flush is about price reaching a cluster of liquidation prices, not about the funding rate being hot. That distinction is the whole point: the fuel for a "grab" is positioning geography, not sentiment.

This is also why the same feed sat at the 86th percentile of its trailing month for liquidation intensity while nothing dramatic was in the headlines. Forced selling is a background process, not a rare event.

The part that is myth: nobody is hunting your stop

Now the expensive half.

The stop-hunt story usually carries two claims that the data does not support.

Claim one: someone is targeting your stop specifically. They are not, because they cannot see it. A stop-loss is a conditional instruction that lives on the exchange and becomes an order only when triggered. It is not visible in the order book beforehand. What large participants can see and model is where stops and liquidations are likely to be clustered, and that is not a secret about you, it is arithmetic about everyone. Equal highs, round numbers, the underside of an obvious range: those are the places leverage stacks, so those are the places that get tested. It feels personal because you put your stop in the obvious place, and so did ten thousand other people. The market went where the orders were, not where you were.

Claim two: a grab reliably reverses price. This is the belief that turns an observation into a bad trade. "Wait for the liquidity grab, then fade it" only works if the sweep is systematically followed by a reversal. We tried to measure that on our own data and the honest answer is that we cannot confirm it yet. Conditioning on sharp one-sided taker sweeps and on abrupt deleveraging over our 25-day state archive, every forward-return horizon came back with too few independent episodes to measure, what our engine flatly reports as history silent rather than dressing a tiny sample up as a signal. A sweep is sometimes the low and sometimes the first leg of a trend that keeps going. The chart remembers the ones that reversed and quietly forgets the ones that did not.

For the version of this question we can answer with years of data rather than weeks, see the 789-day study of forward returns after extreme funding: even there, the popular "crowded side always flushes and reverses" story held on only two of six assets. Reversals after positioning extremes are real sometimes, on some markets, and never the free money the folklore implies.

Sweep, grab, stop hunt: which word is right?

Practically, treat them as one event with three levels of editorializing.

Say sweep when you just mean price took out a level's resting orders. It commits you to nothing about what happens next. Say grab or stop hunt when you want to imply intent and an expected reversal, and know that you are adding a prediction the word smuggles in for free. The discipline is to describe the sweep first, and only call it a grab after price has actually reversed, not before. Most losing "stop hunt" trades are someone naming the reversal before it exists.

How to actually see one

If you want to trade around these events instead of getting run over by them, watch the forced flow directly rather than drawing zones after the fact.

The liquidation tape. Size and side, cross-exchange. A genuine cascade shows up as a burst of one-sided liquidations, hundreds of events in minutes, concentrated on one side. That is the sweep happening. When the burst is long liquidations, downside is being force-fed; when it flips to shorts, the fuel on that side is spent. The liquidation feed and the heatmap explainer both show this live.

Order-book imbalance. Before a sweep, the book often thins on the side price is about to run into, so a modest push travels a long way. Imbalance is the positioning view; a book that empties ahead of a level is a book that is easy to sweep. The walls that flicker in and out around those same levels are a related game: spoofing and fake liquidity.

CVD, the aggression tell. Cumulative volume delta separates a sweep driven by real aggressive market orders from one that is mostly liquidations feeding on themselves. If price took out the level but CVD barely moved, the "grab" was thin, forced flow with no conviction behind it, which is the setup most likely to snap back. If CVD ran hard in the direction of the break, aggressive traders were leaning into it, and "fade the grab" is fighting real flow.

None of these three tells you the future. Together they tell you what actually happened at the level, which is more than a drawn rectangle does.

What this doesn't tell you

The liquidation feed covers Binance, Bybit and OKX, not every venue, so treat the dollar totals as a large representative sample rather than the whole market. The forward-return test is limited by a young state archive; the funding study it points to is the stronger version and still refuses to promise reversals. And the live numbers in this piece are a snapshot from one afternoon, included to show the mechanism, not to characterize a regime. Pull the current figures yourself before drawing any conclusion about right now.

History, not advice. MarketTrace measures what the tape did. What you do with a cascade in progress is your call.

Frequently asked questions

What is a liquidity grab in trading?

A liquidity grab is a fast move that pushes price into a zone where stop-loss and liquidation orders are clustered, triggers them, and uses that forced flow to extend the move. The orders in the zone are the "liquidity"; grabbing them means filling them. It is a real, mechanical event visible on the liquidation tape, where a genuine sweep prints as a burst of one-sided liquidations across exchanges.

Are stop hunts real, or is it a cope?

Both halves have a true part. Stop hunts are real in the sense that clustered stops and liquidations do get triggered in fast moves, and the forced orders really do extend the move, which is measurable. They are a cope in the sense that no one is targeting your individual stop (it is not visible until it fires) and a "hunt" does not reliably reverse price. The event is real; the conspiracy and the guaranteed bounce are not.

What is the difference between a liquidity sweep and a liquidity grab?

They describe the same price action. "Sweep" is neutral: price took out the resting orders at a level. "Grab" (and "stop hunt") add an implied intent and an expected reversal. Using "sweep" keeps you honest, because it does not pre-commit you to a bounce that may not come.

Can you actually see a liquidity grab or stop hunt happen?

Yes, on the forced-flow data rather than the candles. A cascade shows as a cluster of one-sided liquidations in a short window; order-book imbalance shows the book thinning ahead of the level; and CVD shows whether real aggressive orders drove the sweep or it was mostly self-feeding liquidations. MarketTrace shows all three live across Binance, Bybit, OKX and Hyperliquid.

Does a liquidity grab mean price will reverse?

Not reliably. A sweep is sometimes the exact low or high and sometimes the first leg of a continuation. On MarketTrace's own history the "sweep then reverse" pattern does not yet reach a measurable base rate, and the multi-year study of reversals after positioning extremes found they held on only some assets. Treat a reversal as one possibility to confirm with flow, not a rule.

---

MarketTrace shows aggregated funding, order-book imbalance, CVD, open interest and cross-exchange liquidations for BTC, ETH, SOL, BNB, XRP and DOGE across Binance, Bybit and Hyperliquid (liquidations cover Binance, Bybit and OKX). Informational data feed only. Not financial advice.