Where the Stops Actually Live: A Data-First Look at Liquidation Cascades

Every trading floor has its ghost stories. On ours, the most persistent one is the stop hunt: the belief that some invisible hand — the algos, the market makers, the big funds — deliberately pushes price through obvious stop clusters, fills at your pain threshold, and then reverses course, leaving you liquidated and shaken.

It's a satisfying story. It converts bad luck into villainy. But satisfaction isn't evidence, so three members of the collective spent a quarter pulling stop-run events from ES, NQ, CL, and NG — several thousand episodes where price swept a recent swing high or low by more than a defined threshold and then moved at least one more threshold in the original direction or reversed sharply.

What we found won't comfort the conspiracy-minded, but it will make your stops harder to run over.

The Liquidity Vacancy, Not the Villain

The cleanest finding from our sample: the overwhelming majority of what traders call a stop hunt is better described as a liquidity vacancy. Stop clusters form at swing highs and lows because that's where retail and systematic traders alike place protective orders — visible, symmetric, predictable. When price approaches those levels, the resting stops convert into market orders. Market orders consume book depth. If the book is thin at that price — and it usually is, because passive liquidity providers pull quotes ahead of obvious stop shelves — the fills cascade.

Nobody needs to "push" price. The stops push themselves. In our data, roughly four out of five sweep events showed order book depth on the dominant exchange thinning by more than half in the ten seconds before the sweep. The liquidity didn't get hunted. It left the building before the wolf arrived.

This distinction isn't academic. If a villain pushes price, your defense is hiding your stops better. If a vacancy consumes price, your defense is understanding where depth actually rests — and placing stops where the market can absorb your exit rather than where it can't.

Reversal or Continuation: The Honest Split

The folklore says sweeps reverse. Our sample says: sometimes. Among events where price swept a prior extreme by more than 0.15% in the index futures, about 45% reversed meaningfully within thirty minutes and 55% continued. In crude, the continuation skew was sharper — energy futures swept through highs and lows and kept going more often than not, particularly during overnight hours when book depth in CL rivals a ghost town.

The variable that best predicted reversal wasn't time of day or sweep size. It was restoration speed: how quickly depth returned to the swept level after the cascade. When the book refilled within sixty seconds, reversal odds climbed well above the base rate. When the book stayed hollow, continuation dominated. A sweep that empties the shelf is a breakout wearing a disguise. A sweep that gets instantly re-stocked is exhaustion.

A Practical Framework for Stop Placement

We're not going to pretend a dataset turns stop placement into a science. But three rules fell out of the work, and they've changed how several of us manage risk.

First, stop beyond the vacancy, not beyond the level. If you're long ES and your invalidation sits at a swing low, recognize that the swing low itself is a liquidity desert. Place your stop where depth reappears — typically the prior consolidation shelf, not the wick. Yes, that costs more ticks of risk. Size down accordingly. Paying for space in a liquid neighborhood beats dying cheap in a vacant one.

Second, treat obvious extremes as execution zones, not risk zones. If the sweep-and-restore pattern holds — depth returns fast, price reclaims the level — that's information. Several members now use stop-run events as entries in the reversal direction rather than fearing them as exits. The cascade creates the very dislocation mean reversion feeds on.

Third, respect the energy caveat. The continuation skew in CL and NG means the classic "sweep and rip your face back" trade is materially less reliable in energy than in index futures. Overnight energy stops should assume the sweep is the beginning of a move, not the end of one. Our crude postmortem from last year said much the same thing from the loss column; the data now says it from the sample column.

Stop Complaining About the Hunt

The uncomfortable conclusion of the study is that the stop hunt narrative persists because it flatters us. It says our losses were stolen rather than misplaced. The tape says something less flattering: we cluster our risk at the most predictable prices on the chart, in the thinnest sections of the book, and then act surprised when the machinery of the market does what machinery does.

The traders who last in this business don't hide from liquidity dynamics. They map them. Know where the depth rests, know when it vanishes, and put your stop where the market can actually fill you without a fight. That's not paranoia. That's placement.