Expansion: Volatility Has No Direction
Expansion is a stretch in which the distance price covers per bar increases. It is the second half of the contraction-expansion cycle, and that alternation is the well-established part. What it cannot tell you is direction, because a measure of distance carries no sign.
How it works
Expansion is a period in which the distance price travels per bar increases. The market covers more ground in the same time. Nothing in that mentions where it is covered.
It is the other half of contraction. Contraction is the quiet phase and expansion the active one, and a Bollinger squeeze is simply a name for the first of the two.
Quiet and active alternate, which is the whole idea. Volatility clustering — quiet stretches tending to precede active ones — is one of the few claims here with real evidence behind it.
The gap between the two states is large enough to matter. On this site’s shared 576-bar history the 14-bar average true range has a median of 0.5994, a tenth percentile of 0.2823 and a ninetieth of 0.7954 — a ratio of 2.82.
Bar by bar the spread is wider still. Ranges run 0.17 to 1.101 between the tenth and ninetieth percentiles, and the largest bar here is 2.338 to a smallest of 0.022.
Size without a sign
It says nothing about which way it expands. The range opening up is a statement about how far price is travelling, and distance has no direction attached to it.
Because volatility has no sign. The measures underneath it — average true range, historical volatility, band width — are built from absolute or squared distances, which discard the direction of every move.
Participation usually expands with it. When volume rises as the range opens up, more people are trading than in the quiet stretch. A real mechanism, and still no direction.
In practice
A slower chart smooths the whole cycle away. Aggregate enough bars and quiet and active stretches average into each other, so the cycle you see depends on the interval you chose.
And expansion often starts with a gap. Conditions change overnight, so the opening gap does the first leg and the entry you planned never trades at the price you planned it.
So a fixed stop is the wrong size on both sides. A stop loss set at a habitual number of points is too tight once the range opens up and too loose while it is closed.
Costs stay fixed while the range does not. Round-trip cost here is 0.0098 price units — 2% of a median bar’s range, but 45% of the smallest bar. Trading is dearest in real terms when the market is quiet.
It is a description of the past fourteen bars. A volatility reading is a backward window: the conditions you have been trading in, not the ones arriving.
Turning a volatility reading into a size
Start with the money, not the chart. Decide what a single loss is allowed to cost — that is your risk per trade, fixed as a share of the account and settled before you look at a price.
Then turn the current volatility reading into a distance. Multiply the latest ATR by whatever multiple your plan uses, and that product is the stop distance in price units for this trade only.
Position size is then division, not judgement. Divide the money you are risking by the stop distance, and the answer is how many units you can hold. Round down.
The point is that the size moves when conditions do. Because the stop distance is read fresh each time, an expanding market produces a smaller position automatically and a quiet one a larger. You never decide to cut size; the arithmetic does it for you.
What expansion is not
- Not a directional signal. It measures distance travelled, and distance has no sign.
- Not a forecast. The reading describes bars that have already printed.
- Not an indicator. It is a market condition; Bollinger Bands, Keltner channels and the TTM squeeze are ways of seeing it.
- Not a reason to hold more. A wider range means the same size carries more risk, not less.
When it fails
- In a trading range the expansion ends back inside it. Volatility rises, price leaves the edge, and the market carries on between the same two levels. The reading was right; the trade was not.
- When the break that started it does not hold. On this series a breakout above a 20-bar high closed back below the level within ten bars in 85% of 39 events, and above a 55-bar high in 100% of 11 events. Right about the expansion and wrong about the direction is the ordinary outcome, which makes a false breakout the base case.
- When you were sized for the quiet phase. A position set during contraction carries the new range whether or not you re-examined it.
- When the expansion is one bar wide. A single wide bar prints on a headline and leaves nothing behind, so the condition ends before an entry can be placed.
- When you drop to a faster chart to catch it earlier. Costs stay fixed while bar ranges shrink, and round-trip cost already exceeds 10% of a bar’s range on 15 of the 576 bars here.
- When distance travelled is not distance kept. The ten-bar efficiency ratio has a median of 0.34, with 30% of bars above 0.5, and the typical stretch gives back most of its ground.
The original data
The two ratios in research/series-measurements.json are the point. Smoothed over fourteen bars
the quiet-to-active ratio is 2.82; bar to bar it is 6.5 between the tenth and ninetieth percentiles,
and 106 to 1 between largest and smallest. A size suiting a median bar is roughly three times too
large in quiet conditions and too small in active ones.
Costs are fixed while the range is not. The same round trip is 0.0098 price units, 2% of a median
bar and 45% of the smallest, so the quiet phase is the expensive one. research/broker-coverage.json,
scanning 31,760 videos in research/search-study-corpus.jsonl, finds “expansion” in 10 titles,
median 142,645 views, 8 channels, maximum 1,150,563. Before the next trade, take the current average
true range reading from site/measure_series.py and size from that, not from a fixed number of
points chosen when the market was calm.
Related
ATR is the reading that turns this page into a number you can size from.
Bollinger squeeze is the contraction half of the same cycle, and the condition that usually precedes expansion.
Breakout is the event expansion arrives with, and the page separating a break that holds from one that does not.
The mistake I made for years was keeping my position size the same while the market changed underneath it. I would settle on a size during a quiet stretch, and then the range opened up and that same size was carrying far more risk than I had agreed to. Nothing in my process had changed, which is exactly why I could not see it. Now the size is calculated from the current volatility reading every time, and I do not get to overrule it.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.