What Is the CCI Indicator?
CCI, the commodity channel index, measures how far the typical price is from its own moving average, divided by the average deviation over the same window. It is unbounded, so the ±100 lines are a convention rather than a limit.
The name says commodity and the tool has nothing to do with commodities. What it actually measures is distance, and the interesting part is what it is divided by.
How it works
Three steps:
typical price = (high + low + close) / 3
CCI = (typical price − its 20-bar average) / (0.015 × mean deviation)
The numerator is the distance from average. The denominator is how far this instrument normally strays. So the output is a distance measured in units of its own usual distance.
The 0.015 is a scaling constant chosen by Donald Lambert in 1980 so that roughly 70–80% of readings would land inside ±100. It has no meaning beyond making the numbers a convenient size.
It has no ceiling
This is the real difference from RSI and the stochastic, and it changes how you read it.
On this chart CCI ran from −246 to 317. RSI on the same data cannot leave 0–100 no matter what price does, because its formula divides by its own total.
So “extreme” means something different here. A bounded oscillator pins at its limit and stops telling you anything; CCI keeps counting.
And ±100 is not a limit, it is a line someone drew. Price passed 100 on 12 of 54 bars here — about one bar in four and a half.
The zero line
Zero means price is exactly at its 20-bar average. Above zero it is above average; below, below.
That makes a zero cross the same event as a price/moving-average cross — which it is, arithmetically. It was crossed 11 times on this chart, so treating each one as a signal produces eleven trades on one screen.
What it is really measuring
The same thing Bollinger Bands measure.
A Bollinger band asks how many standard deviations price is from its 20-bar average and draws the answer as a line. CCI asks how many mean deviations price is from its 20-bar average and writes the answer as a number.
Different dispersion measure, same question. So running both and treating agreement as confirmation is counting one measurement twice — the objection the technical analysis page makes about derived indicators, in a form that catches almost everybody.
The settings
One number: the lookback, 20 by default.
Measured on this chart, length 20 and length 10 both put 23 bars beyond ±100. The shorter one is visibly twitchier and produced the same count — which is a useful reminder that “more sensitive” and “more signals” are not the same thing, and neither is “better”.
Mean deviation, and why it matters
The denominator is the part that makes CCI different from every other oscillator here, and it is worth one section.
Bollinger Bands divide by the standard deviation. CCI divides by the mean deviation — the average of the absolute distances from the average, with no squaring.
Squaring is what makes standard deviation react to outliers. A single enormous bar is squared before it goes into the average, so it dominates the result; in mean deviation it counts once.
The practical consequence: after one violent bar, a Bollinger band widens sharply and CCI does not. So on the bar after a shock, the two disagree — the band says conditions are extreme, CCI says price is a long way from average. Both are correct about different questions, and knowing which one you asked is the difference between a reading and a guess.
That is also why the two are worth having as a pair only if you know that is the distinction you are buying. Otherwise it is one measurement, twice.
A worked example
Establish the condition first. CCI is a distance-from-average measure, so it is a trading range tool. In a trend it will read extreme for weeks.
In the range above it stayed between −103 and 135 — the ±100 lines land close to the actual turning points, which is the condition Lambert designed it for.
Mark the range edges from price, then use a reading beyond ±100 near an edge as a second reason.
The invalidation is a close outside the range, from the chart. CCI has no price in it and cannot give you one.
The original data
Across our study of 24,971 trading videos, 69 cover CCI. The median one gets 12,334 views, 75% never pass 50,000, and the median length is 9.6 minutes.
The corpus carries description text for 63 of those 69 — an unusually complete sample — and across those 63, zero mention invalidation, failure, or what a bad read looks like.
Zero out of sixty-three, on an unbounded oscillator whose most common use is treating a bound it does not have as a signal.
When it fails
In a trend it stays extreme
The same trap as every oscillator here, and worse for one with no ceiling: there is no level at which it stops going up, so “it cannot get more overbought” is never true.
Selling the first reading over 100 in the advance above would have been wrong, and so would every reading after it.
±100 is arbitrary and gets treated as physics
Covered above. The 0.015 constant exists to make about three quarters of readings fall inside ±100, and that is the entire basis of the threshold. Nothing about a market changes at 100.
It duplicates a band you may already have
If Bollinger Bands are on the chart, CCI is the same measurement in a panel. Two views of one number is not two opinions.
You read the extreme after price turned
Every top has a high CCI reading before it. So does every continuation — and with no ceiling, “very high” carries no more information than “high” did.
Related
Bollinger Bands is the same measurement drawn on price, and the comparison is the fastest way to understand either.
Moving average is the thing CCI measures distance from.
And RSI is the bounded alternative — worth reading alongside, because the difference between bounded and unbounded is the whole of this page.
What made this click for me was realising it is asking the same question as a Bollinger band, just answering it with a number instead of a line. Both are measuring how far price is from its average in units of how far it usually goes. Once I saw that I stopped running both, because they were telling me the same thing twice and I was counting it as agreement.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.