ICT Concepts: How Many Setups Are on a Chart?
Inner Circle Trader (ICT) concepts are a set of named chart events - imbalance, order block, liquidity sweep, break of structure and others - each defined precisely enough to apply mechanically. Applied that way they label a large share of any chart, so selectivity has to come from the trader rather than the names.
The concepts are usually taught one at a time on a chart chosen to show them. This page applies four of them mechanically to the same 144 bars and counts what comes out.
How it works
Each concept is a rule about bars. Written exactly, so a computer could find them and so can you:
Imbalance — a three-bar gap: this bar’s low is above the high two bars back, or its high is below the low two bars back. Also called a fair value gap.
Order block — a down candle followed within three bars by a move of more than 1.5 median ranges. The order block page covers what it is meant to represent.
Liquidity sweep — a bar that trades below a prior swing low and closes back above it. The liquidity sweep page has the mechanism.
Break of structure — a close above the prior swing high, from the market structure definition.
Those four definitions are the genuinely valuable part of the vocabulary. A concept you can code is a concept you can count, and counting is what the rest of this page does.
How much of a chart they label
Applied to 144 ordinary bars, the four rules produced 84 labels across 65 different bars.
| Concept | Bars |
|---|---|
| Imbalance | 29 |
| Break of structure | 25 |
| Order block | 17 |
| Liquidity sweep | 13 |
45% of the chart qualifies for at least one name, and that is with only four concepts. The full vocabulary has several times as many.
This is not a criticism of the definitions. They are doing exactly what they say. It is a statement about what naming can and cannot do for you: a label that applies to nearly half the bars is not a filter.
The same bar, several names
19 of the 65 labelled bars carried more than one name.
The most common pairing was imbalance with break of structure — nine bars. Which makes sense: a bar that closes above the prior swing high has usually moved far enough to leave a three-bar gap behind it.
So some of these concepts are not independent observations. They are different measurements of the same event, and a setup described as “an imbalance at a break of structure” is often one thing counted twice.
The choosing indicators page makes the identical argument about oscillators, with correlations instead of counts. Confluence between two things that measure the same event is not confluence.
The older names
Most of these events had names before. An order block is the last opposing candle before an impulse, which a supply and demand trader calls a zone. A liquidity sweep is what Wyckoff called a spring in the 1930s.
A change of character is the first lower low in an uptrend — a structure break, described on the reversals page without the acronym.
None of that makes the newer names worthless. A shared vocabulary is genuinely useful, and the Inner Circle Trader (ICT) versions are usually stated more precisely than the older ones. What it does mean is that learning them is learning names, not learning new mechanisms — and the ICT page sets out the three parts that really are its own.
A worked example
Pick two concepts, not ten. Write each one down as a rule a stranger could apply.
Count how often each fires on your own chart across a few hundred bars. If it fires on a third of the bars, it is describing the market rather than selecting from it.
Check whether your two fire on the same bars. If they do, you have one.
Then decide what else has to be true — the direction, the level, the session — because that is where the selectivity actually comes from.
The original data
Across our study of 24,971 trading videos, 1,192 cover ICT concepts — the third-largest field measured anywhere in this glossary, behind only forex at 1,639 and day trading at 1,465. The median one gets 9,738 views, 70% never pass 50,000, and the median length is 15.7 minutes.
The corpus carries description text for 371 of those 1,192, and across those 371, 17 mention invalidation, failure, or what a bad read looks like.
4.6%, on the largest body of teaching material in the study. The median length of 15.7 minutes is also well above the site-wide norm, so these are long videos: a great deal of explanation, and one description in twenty-two mentioning a case where the read was wrong.
When it fails
The names become the analysis
In a flat, directionless stretch, 15 of 42 bars still qualified for one of the four names.
A chart with nothing happening on it is still full of labels, which is exactly when a vocabulary that labels everything is most dangerous.
Confluence is counted twice
Two names on one bar is one event. Treating it as two agreeing signals is the confluence error, and it is easy to make when the concepts overlap by construction.
The definition drifts
“Order block” applied loosely will fit almost any candle before a move. The definitions above are narrow on purpose — the moment one is applied by feel, the count stops meaning anything and so does the setup.
You found the label after the move
Every one of the 65 looked like this at the time. Naming an event after seeing what followed it is the whole hindsight problem, and a rich vocabulary makes it easier rather than harder.
Related
ICT trading is the methodology, and the three parts of it that are genuinely original.
Order blocks and market structure are two of the four definitions above, each with its own mechanism.
And choosing indicators is the same overlap argument, measured with correlations.
I use two of these and ignore the rest, and that is not a judgment about the others. It is that the number of things I can genuinely watch for is small, and a vocabulary that names forty events is a vocabulary in which I will always find one that fits what I already wanted to do.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.