Kijun-sen: The Flat Parts Are the Point
The kijun-sen is the base line of the Ichimoku system, calculated as the highest high and the lowest low of the last twenty-six bars, divided by two. It is a midpoint of extremes rather than an average of closes, so it sits perfectly flat whenever neither extreme changes.
How it works
The kijun-sen is a midpoint of extremes, not an average of closes. Take the highest high of the last twenty-six bars, add the lowest low of the same twenty-six bars, and divide by two.
It is the same calculation the tenkan-sen runs, over a longer window. The conversion line looks back nine bars, the base line twenty-six. The conventional Ichimoku settings are nine, twenty-six and fifty-two.
A longer window means a slower, later line. A bar has to beat twenty-six bars of extremes to shift it, not nine. That delay is the trade-off.
Why it goes flat
Because only two numbers feed it, the line freezes whenever neither changes. Price can wander inside the existing twenty-six-bar range for weeks without touching either extreme, and the output stays identical.
Those flat stretches are what traders actually watch. A flat base line marks a stable range midpoint, and it is the most legible thing the system draws.
It is an equilibrium price, not a trend line. The calculation says where the middle of the recent range sits, and nothing about direction.
In practice
Price crossing the base line is the slower of the system’s two crossing signals. The conversion line crosses first and more often. Later is not better.
Like every price-derived line, it ignores participation. No amount of volume behind a move changes the arithmetic, because volume is not an input.
Twenty-six daily bars is roughly a trading month, and that is no accident. The nine, twenty-six and fifty-two settings come from a six-day trading week that no longer exists.
An opening gap that sets a new extreme moves the line in one step. A smoothed average would absorb that jump over several bars; a midpoint of extremes cannot.
Its most defensible use is as a trailing stop reference on an open position. You have the trade already; the line only says whether the range midpoint moved with you.
Every exit on it pays the same friction as any other. A round trip on this site’s shared series costs a measured share of one bar’s range, and a stop on a flat line is hit alongside everyone else’s.
Nobody is defending the line, it is a computed midpoint. No resting size sits behind an arithmetic result — only the attention of everyone running the formula.
Why the flat levels hold at all
When a flat base line holds, the mechanism is self-fulfilling rather than structural. Nothing in the order book is anchored to the midpoint of twenty-six-bar extremes. What is true is that many people watch the same arithmetic on the same chart.
That distinction matters because self-fulfilling levels fail differently from structural ones. A level built from real accumulation has size behind it. A shared computation has only shared attention, which evaporates once enough participants stop looking.
It also explains why the effect is strongest on liquid, widely charted instruments. The more traders running default Ichimoku settings on a symbol, the more its flat sections behave like levels. Thin instruments show little of it.
So treat a reaction at the line as a coincidence of attention, not evidence of defence. It is worth watching for that reason, and worth never overstating for the same one.
What the kijun-sen is not
- It is not a moving average. It uses two extremes, not every close in the window.
- It is not a trend indicator. A midpoint reports a range’s centre, never its direction.
- It is not a support level. No resting orders defend a computed number.
- It is not an entry system. It is a reference for a position you hold.
When it fails
- In a range it sits flat in the middle and catches everything. Price crosses it repeatedly in both directions and none of those crossings resolves.
- After a gap it is already stale. The line reprices in one step, so a stop placed against yesterday’s level is now measured from somewhere else entirely.
- In a fast trend it trails a long way behind. Twenty-six bars of lookback puts the midpoint far under a strong advance, giving back a great deal before the line registers a turn.
- On thin instruments one spike distorts it for twenty-six bars. A single bad print sets an extreme, and that extreme rules the calculation until it rolls out of the window.
- Once the flat level is widely published it becomes a target. Visible clusters of orders attract the participants who trade against them.
- It says nothing about participation. A collapse on heavy volume and a quiet drift on almost none produce exactly the same line, and the chart gives you no way to tell them apart.
The original data
Two measured files sit behind this page. research/corpus-coverage.json, from
site/measure_corpus.py, covers 31,760 trading and investing videos.
research/series-measurements.json, from site/measure_series.py, measures this site’s shared
576-bar history.
- Title coverage in the corpus. “Kijun”: 5 videos, median 33,420 views, 3 channels, maximum 116,694. “Tenkan”: 2 videos, median 116,694 views, 2 channels. “Kumo”: 5 videos, median 17,983 views, 3 channels. “Ichimoku”: 154 videos, median 10,165 views, 99 channels.
- Round-trip cost. 2% of a median bar’s range and 45% of the smallest bar.
- Trailing-stop survival. Opened at every eligible bar and trailed by a multiple of the 14-bar average true range, the median position survived 3 bars at one average range, 10 at two, 22 at three and 32 at four, across 562 trials each; between 91% and 100% were eventually stopped out.
- Direction runs. Average 2.01 bars, longest 11, across 286 runs.
The corpus numbers describe attention, not merit. Ichimoku is covered by 99 channels and the base line by three, so the component is discussed inside a system rather than alone. That is roughly how it deserves to be used.
The survival table is the part that should change your behaviour. Widening the trail from one average range to four lifted median holding time from 3 bars to 32, yet nearly every position was still stopped out — a wider trail buys time, not immunity. Before trailing on this line, write down the number of bars you expect to hold and check it against the table.
Related
Ichimoku is the parent framework. The base line is one of its five plotted components. None of them makes much sense alone, so read that page first.
The tenkan-sen is the same formula over nine bars. It turns faster and goes flat less often. Comparing the two shows what the lookback setting buys.
The kumo cloud is built partly from the base line, projected forward. It carries the system’s only forward-looking claim. Read it after this page, not before.
I do not take entries off this line, and I stopped trying a long time ago. What I do is leave it on the chart as something to trail against once a position is already open and already working. It answers one narrow question well: has the middle of the recent range moved in my favour or not? When it stops moving, that tells me something about the market before it tells me anything about my trade.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.