Back to Guides
Guide

What Is a Volatility Contraction Pattern? The Numbers

What is a volatility contraction pattern? Minervini's shrinking-pullback base, the four depth sequences, and the one test that read literally never fires.

A volatility contraction pattern is a base inside an uptrend where each pullback is shallower than the one before it, volume dries up as it goes, and the whole thing ends in a tight area whose high becomes the level to watch. It is Mark Minervini's name for the shape, and the reasoning is absorption of supply. Sellers who wanted out around these prices have mostly gone, so each dip meets less selling than the one before it.

The numbers are the weak part. Four different depth sequences are in circulation, the steps they imply run from roughly a half to two thirds, Deepvue contradicts itself inside its own pages, and a test written as a volatility measure rather than a pullback depth, TradingSim's, read literally never fires at all on nearly three years of our own data. There is no canonical set of percentages to copy, and the loosest published step ratios cannot reach the tightest published final depth in only three contractions.

What does a volatility contraction pattern look like in numbers?

The published sequences, with the ratio each step implies:

SourceSequenceStep ratios
LuxAlgo20% then 10% then 5%0.50, 0.50
Deepvue15% then 10% then 5%0.667, 0.50
tradingmomentum25% then 15% then 7%0.60, 0.467
AIO Indicator22.5% then 12.5% then 6.5% then 4%0.556, 0.52, 0.615

The last row is a midpoint reading: AIO Indicator gives a band for each contraction rather than a single figure, so the row is the middle of each band.

LuxAlgo's step rule says each contraction should be "commonly near half its depth". Across every published step above, the real spread runs from 0.467 to 0.667, and the only sequence that obeys "near half" is the tidy 20-10-5 illustration LuxAlgo states it beside.

Deepvue is looser than that, giving two sequences on its own pages that disagree with each other: a worked example that goes from a 15% pullback straight to a 5% one, and a bullet list a few paragraphs later that puts a 10% step in between. That is a ratio of 0.333 in one place and 0.667 in another, from the same publisher.

The percentages are best read as a sketch of a shape. The tradingmomentum Substack says so outright: "The exact percentages and contractions do not need to be perfect."

How many contractions does a VCP need?

Two to four is the usual answer, with five or six as the exception, and the published counts nearly agree. LuxAlgo says two to four is typical and five or six happens. The tradingmomentum Substack says sometimes only two or three, and other times as many as five or six. TradingSim asks for at least two to four. AIO Indicator calls two to six typical. Deepvue gives no count at all.

The problem is that the published depths and the published counts cannot both be right at the published ratios. Taking AIO Indicator's final contraction of 3% to 5% as the target, which is the tightest published band, and stepping a constant ratio down from each published starting depth, the sequence first lands in that band at:

Step ratioFrom a 20% first pullbackFrom a 25% first pullback
0.50C3C4
0.556C4C4
0.60C4C5
0.667C5C5

At two thirds you need five contractions from either starting point, and from 25% so does a 0.6 ratio. Three of the four sources that give a count allow five or six, so a base that long is inside what they describe. The clash is between the loosest ratios and the low end of those same ranges: at two thirds, a base of three or four contractions never gets tight enough to qualify.

Pick the count or pick the depths, and let the other one go. If you are going to insist on a 3% to 5% final contraction, you need either a shallow first pullback or a ratio near a half, and if you are going to insist on three contractions you have to let the final one be deeper than 3%.

How do you identify a volatility contraction pattern?

Five things, and only the last of them depends on what price does next.

  1. An uptrend before the base. The pattern is a continuation shape, not a bottom.
  2. A run of pullbacks, each one shallower than the last by a clear margin. AIO Indicator makes this the hard test: "each contraction must be smaller than the one before it." If the second is deeper than the first, volatility is expanding and the setup is finished.
  3. Higher lows through the base. Each pullback bottoming above the previous one is what distinguishes absorption from a structure that is deteriorating.
  4. Volume falling away as the base tightens, and visibly thin by the last contraction.
  5. A pivot, which is the top of that last tight stretch. Deepvue and the tradingmomentum Substack both call it the line of least resistance, and the shape is only finished when price pushes through it with volume expanding. TradingSim's breakout filter asks for volume "at least 40 to 50 percent above average volume on the breakout day".

LuxAlgo is the one that writes all of that as a specification: the pivot is the highest high of the final contraction, volume must sit below its own moving average all the way through that contraction, and the trigger is a close above the pivot with volume back above its moving average.

Sideways chop is what the pattern gets confused with, and no arithmetic test separates the two. The tradingmomentum Substack puts it as a base where "the structure becomes more controlled as the base develops", while chop keeps swinging hard and cutting back under the lows it has already made. LuxAlgo, which wrote that specification, also warns that tightness buys you less than it looks: "Tightness improves the risk-reward framing; it does not make the breakout succeed."

How tight does the final contraction have to be?

TradingSim answers with a volatility measure rather than a pullback depth, and its sentence is ambiguous. It says price must compress "into a narrow range where the average true range falls to roughly one-third of its 50-day average", and that has two readings which are nothing alike.

Reading one is literal: a 14-day average true range at or below a third of its own 50-day average. Reading two is looser: a single bar's own range at or below a third of the 50-day average range.

We measured both on daily bars built from our cleaned one-minute history, on EUR/USD and on the German index, from the start of 2024 to the end of August 2026.

EUR/USDGerman index
Days measured835832
Reading one, days qualifying0 of 7720 of 769
Quietest day, as a multiple of the threshold2.0x1.4x
Reading two, days qualifying90 of 785 (11.5%)88 of 782 (11.3%)
Runs of three or more such days0, longest run 11, longest run 4

Both readings lose days to warm-up before their averages produce a first value, which is why each one qualifies out of fewer days than the days measured, and reading one, an average of an average, loses the most.

Reading one never happens: not once in nearly three years on either instrument, and the quietest day on record is still double the threshold on one and about half again on the other. A 14-day average cannot fall to a third of its own 50-day average without a volatility collapse neither instrument produced.

Reading two is ordinary, firing on about one day in nine. But AIO Indicator describes a run of three to five tight, quiet days as something you often see rather than something the pattern requires, and runs of three or more never happened at all on EUR/USD and happened once on the index.

If you are quoting that one-third figure, say which reading you mean, because one of them is unreachable. And these are forex and an index, not stocks, so read them as evidence about how strict the condition is rather than as a base rate for equities, where volatility behaves differently.

Is a volatility contraction pattern the same as a narrow range bar?

No, and they are not even the same size of object. A narrow range bar flag such as NR4 or NR7 is a property of one bar, whose range is the smallest of the last four or seven. A volatility contraction pattern is a multi-week base made of several pullbacks and recoveries.

A tight bar can appear inside the final contraction of a base, which is probably why the two get mixed up, but a rule written for one will not find the other.

How do you test a VCP by hand?

Write the whole thing down before you open a chart, and be more specific than the published material is. How many contractions you require, the depth of each one as a percentage of the prior high, the ratio test, whether higher lows are mandatory, where the pivot sits, the volume condition through the base, and the volume condition on the breakout. Then step forward one bar at a time with the next bar hidden, which is the only way to mark a pivot before you know whether it broke.

The limit is what you can test here. The geometry travels: the depth sequence, the ratios, the pivot level and how price behaves when it gets there. The half that makes it Minervini's method does not. Relative strength ratings against a market index, quarterly earnings growth and share-supply absorption all need equities, and ChartLabs covers forex, indices, crypto, commodities and metals, not stocks.

Within that limit, bar-by-bar replay is the right tool for a pattern whose whole difficulty is judging tightness without knowing the outcome. The drawing tools have keyboard shortcuts and a saveable default style, which matters when you are marking each contraction's high and low and a pivot line on base after base. Every trade captures a snapshot of your drawings and chart state at the moment of entry. The Orders panel shows up to five tags per trade and carries a notes column, and your notes come through in the CSV export, so there is somewhere to record which ratio rule and which tightness reading a given trade used. Commission and slippage are per symbol, so the result carries its costs. History depth is per symbol too, up to around 22 years on the oldest forex pairs, with other instruments starting later.

There is no screener and no scan. Finding candidates is a different job from testing a rule, and replay is the second one.

Where to go next

Backtesting a Trading Strategy covers the workflow, and Backtesting Accuracy is worth a look before you draw conclusions from a small number of bases, which is what a multi-week pattern on a few years of data will give you.

Then pick one depth sequence, one tightness reading and one instrument, and log every base you can find. ChartLabs is $20 a month or $200 a year, one plan, and the trial is 7 days with no credit card: start the free trial.

Related Topics

volatility-contraction-patternvcpmark-minervinichart-patternsmanual-backtesting