The Volatility Squeeze Set-Up
A market whose range has contracted to the narrowest it has been in months, which says a large move is coming without saying anything at all about its direction.
MadStockAlerts Research · Updated September 4, 2026
What to take away
- This is the one set-up whose shape makes no directional claim; the direction has to come from somewhere else.
- Volatility is the most reliably mean-reverting thing on a chart, which is why the contraction is informative.
- The edges of the squeeze are the two references, and it completes through whichever one goes.
- The first move out of a squeeze is frequently the wrong one, which is a known and expensive failure.
- There is no measured objective; the size of the expansion is scaled to how tight the contraction was.
MAD Academy Training Video · 1:24
The set-up that will not tell you which way
Volatility reverts, so a multi-month low in range says a big move is close. It says nothing about direction — and the head-fake is what happens when that is forgotten.
This lesson is part of a Stock Alerts + Tools plan.
The shape
Ranges narrow, week by week, until the bars are a fraction of the size they were a month earlier. Drawn with volatility bands the two edges close in toward the price and the whole picture pinches, and then one bar breaks out of the range and the bands fly apart.
Unlike every other shape in this catalogue there is no outline to recognise — no flag, no triangle, no ceiling. What is being measured is the width of the recent range against its own history, which means a squeeze can only be identified relative to the market it is in. A range that is tight for one stock is a quiet week for another.
- A range at or near its narrowest reading of the last several months.
- Volatility bands at their tightest, and narrowing rather than merely narrow.
- Volume that has drifted down alongside the range, not spiked.
- A prior trend the expansion could plausibly continue, which is where the direction comes from.
Scroll the chart sideways to see all of it.
- Two standard deviations
Why it forms
Volatility clusters and it reverts. Quiet periods follow quiet periods until they do not, and violent periods follow violent ones until they exhaust. That is one of the better-established regularities in market data and it is the entire foundation of this set-up: a range at a multi-month low is far more likely to widen than to narrow further, because there is very little narrower left to go.
The mechanism is ordinary. A narrow range means the people who wanted to trade at these prices have already done so, and the remaining orders sit further away on both sides. When something finally moves price out of the range, there is nothing immediately above or below to slow it, so the first move is disproportionately large relative to the days that preceded it.
A contraction predicts magnitude, not direction, and treating it as though it predicts both is the mistake this set-up exists to warn about. Everything directional here is imported: the trend the squeeze sits inside, the position of the range within a larger structure, what the sector is doing. The squeeze itself is silent on which way.
| What to look at | Set-up intact | Set-up failing |
|---|---|---|
| The range | Still contracting, at multi-month lows | Widening again with no resolution |
| Volume through the squeeze | Drifting down with the range | Spiking inside the range, repeatedly |
| The surrounding trend | Clear, and the squeeze sits high in it | Absent — a squeeze in the middle of nothing |
The price points that define it
A set-up is a shape plus a handful of prices. The shape is what makes it recognisable; the prices are what make it something that can be measured, reviewed afterwards and argued about honestly. These are the levels this one is read from.
| Level | Where it sits | What it tells you |
|---|---|---|
| The top of the squeeze | The high of the contracted range | One of the two references. It completes through this side if the expansion is upward. |
| The bottom of the squeeze | The low of the contracted range | The other reference, and the level that says the expansion went the other way. Both have to be marked before either breaks. |
| The width of the squeeze | Top minus bottom | The only measurement here. It scales the expansion that follows and it is the number that says whether this is a squeeze at all. |
| The prior trend's direction | The structure the squeeze is sitting inside | Not a price, and the most important input. It is the only thing in the whole set-up that argues for one side over the other. |
Where the set-up completes
- 1The triggerThe set-up completes on a close outside the squeeze accompanied by an expansion in range — a bar noticeably wider than the ones inside it. A close outside the range on a bar the same size as the rest is not an expansion, and it is the single most common false signal this shape produces.
- 2Where the reading stops holdingThe reading stops holding on a close back inside the range. That level is unusually clean here, because the range is narrow by definition: a squeeze that has resolved and then returned inside its own boundaries has un-resolved, whatever it did in between.
- 3The measured objectiveThere is no conventional objective and the shape does not supply one. Some descriptions project a multiple of the squeeze's width, which at least has the merit of being scaled to the thing being measured; none of them has any better claim than that. What the contraction supports is an expectation about size, not a destination.
- 4Through the moveThe convention watches whether volatility keeps expanding. A squeeze that resolves and then immediately quietens again has not really resolved, and the first move out of a long contraction is frequently reversed — which is why many descriptions treat the second move, after a failed first one, as the more reliable of the two.
Each of these describes where a convention puts a level, not what anybody should do at it. Whether a level is worth acting on at all is a question about position size, cost and the rest of a plan, and the answer differs for every account.
How often it follows through
This band describes the shape resolving in the direction of the trend it sits inside, which is the only version in this catalogue. Read as a directionless coin-flip on which side breaks, the number would be far lower, and the gap between those two readings is the whole reason the trend has to be established first.
Read that as a floor rather than an expectation. A set-up that follows through two times in three still leaves one in three that does not, and the one that does not can move further and faster than the two that did. That arithmetic is what position sizing exists to answer, and no pattern improves it.
What this number is not
It is not our record, it is not a forecast for any particular chart, and it is not the rate at which the measured objective is reached — that is always lower. It is a conservative reading of how often a completed volatility squeeze kept going before it went back through the level that invalidates it. Base rates move with the market, the timeframe and the exact definition used, and every one of those varies.
What failure looks like
The characteristic failure has a name in the literature and it deserves one: the head-fake. Price breaks one side of the squeeze, brings in the accounts watching for exactly that, then reverses through the whole range and expands the other way. It happens often enough that some descriptions treat the first break of a long squeeze as information about where the reversal will come from rather than as a signal in its own right.
- The bar that leaves the range is no wider than the bars inside it.
- Price closes back inside the range within a session or two.
- The squeeze is not sitting in any identifiable trend, so nothing argues for either side.
Scroll the chart sideways to see all of it.
- Two standard deviations
Seeing it on a live chart
A squeeze is a ranking rather than a picture: it only means anything measured against the same market's own history. Sorting a list by how narrow the recent range is against the last few months finds them; looking at charts one at a time mostly does not.
Open the Scanner — for members