How to Use the Ichimoku Cloud
To use the Ichimoku cloud, treat the cloud itself as the trend filter and the two fast lines as the timing signal. Two components are shifted forward or back in time, which is why the display looks unusual and why a signal has to be read against the correct bar.
The Ichimoku system draws five lines, two of which are plotted at a different point in time from the bar they were calculated on. That displacement is the reason the chart looks unfamiliar and the reason the system is misread more often than most.
Before you start
A decision about which of the five lines you are actually going to use. Two is a system. Five is a guarantee that something on the chart supports every possible action.
The instrument’s ordinary bar range, because the cloud’s thickness has to be read against something. On this site’s shared series the median bar range is 0.493 with a ninetieth percentile of 1.101.
An acceptance that two of the five components are shifted in time. The cloud is projected forward and one line is plotted backward, which is deliberate rather than a rendering error.
The steps
1. Use the cloud as a directional filter
Price above the cloud means you take long setups only; below it, short setups only. This single rule is most of what the system contributes.
2. Treat price inside the cloud as no-trade
Inside means the components disagree. That is a genuine answer and it can persist for twenty bars, which is a long time to sit out and the correct thing to do.
3. Use the two fast lines for timing only
Their crossover is the entry trigger, taken only in the direction the cloud already permitted. A cross against the cloud is not a signal you act on.
4. Read cloud thickness as disagreement
A thick cloud means the components are far apart, so the region where nothing is decided is wide. A thin one means price can cross the whole zone in a bar or two.
5. Check the displaced line against the right bar
The lagging component is plotted behind the current bar and compares closing price to price some periods ago. Reading it against today’s bar produces a confident wrong answer.
6. Take the stop from structure, not from a line
The far edge of the cloud is sometimes a sensible level and sometimes an arbitrary distance away. Where the idea is wrong is a structural question, and that distance sets the size.
7. Leave the settings alone unless you can say why
The standard periods are widely used, which makes the levels partly self-fulfilling. Changing them is a defensible choice and it removes that property, so it needs a reason you can state.
How to tell it worked
At most 2 of the 5 components have a vote, decided before the chart was opened.
0 trades were taken while price sat inside the cloud.
Every entry was in the direction the cloud permitted, checked before the fast-line cross.
And each stop came from structure, so the size was not set by a line’s position.
What it cannot do
It cannot tell you a move is finished. On this site’s shared series direction runs average 2.01 bars and the longest ran 11, so price sitting above the cloud for a long stretch is ordinary rather than extended.
It cannot see participation. The cloud looks identical on an instrument nobody trades, and every level it draws there is a level nobody is defending.
Why five lines is the problem
Each component is a legitimate calculation. None of them is noise, and each answers a slightly different question about the same price series.
Displayed together they produce a permanent supply of confirmation. In any market condition at least one of the five supports buying and at least one supports selling, which means the chart can be read to agree with any prior intention.
Choosing two in advance is what converts it into a system. The choice is less important than making it before you have a position and writing it down.
The two-component version, written out
Rule one: the cloud sets the permission. Above it, you may look for longs. Below it, shorts. Inside it, you do nothing, and that is a state you should expect to be in for a meaningful share of the time.
Rule two: the fast-line crossover is the trigger, taken only in the permitted direction.
Rule three: the stop comes off the chart, from the level that would prove the setup wrong, and the position size follows from that distance.
Everything else on the display is scenery. The remaining components can stay visible if you find them useful to look at, provided they have no vote — and the test of whether they have a vote is whether you have ever declined a trade because of one.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 151 mention this system in the
title, at a median of 10,245 views across 99 channels, and 68% of those titles are instruction-shaped.
One trend indicator appears in 122 at 19,638 and another in 126 at 3,163. The counts come from
site/corpus_count.py.
151 videos across 99 channels at a 10,245 median. Very wide coverage and a moderate audience per video, which is what happens when a visually distinctive system attracts explanation more than it attracts use.
The answer to the question on that chart is that inside the cloud was already a no. The crossover does not overrule the filter — if it did, the filter would only apply when it agreed with the signal, which is the same as having no filter at all.
When it fails
The failure is the chart with all five lines and no prior decision, and it feels like thoroughness. Every component is on, every one is a real calculation, and in any market at least one of them supports the trade you were already inclined to take. The system never says no, because saying no requires having decided in advance which lines are allowed to speak — and with all five visible, that decision gets made after the setup appears rather than before.
The second failure is trading inside the cloud. That is the indeterminate state.
A third is reading the lagging line against the current bar. It refers to a different one.
A fourth is treating a thick cloud as strength. It is a wide zone of disagreement.
A fifth is taking a stop from a cloud edge. It is a line, not a level.
And a sixth is changing the periods without a reason. Their standardness is doing work.
Related
Ichimoku covers the five components and how each is calculated. Kumo cloud is the projected band and what its thickness means. And Tenkan-sen is the faster of the two timing lines.
I use two of the five components and ignore the rest, which felt like cheating until I noticed that using all five meant I could always find one supporting whatever I already wanted to do. Choosing which lines have a vote, in advance, is the only version of this that produces reviewable decisions.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.