The five lines, written out as arithmetic
Ichimoku uses five plots. Tenkan sen, the conversion line, is the highest high plus the lowest low of the last nine periods, divided by two. Kijun sen, the base line, is that same calculation over twenty six periods. Senkou Span A is the average of those two lines, plotted twenty six periods into the future. Senkou Span B is the midpoint of the fifty two period high and low, also plotted twenty six periods forward. The shaded area between the two spans is the kumo, the cloud. Chikou Span is nothing more than the current close, drawn twenty six periods into the past.
Notice what is absent. There is no exponential weighting, no volume term, no standard deviation. Four of the five plots ignore closing prices completely and use only the extremes of their window. That is a deliberate choice rather than an oversight, because a range midpoint is a crude equilibrium price and the method is built to compare several equilibriums at once. The default numbers come from an era when the Japanese trading week ran six days, which is why nine, twenty six and fifty two look arbitrary today.
Why the cloud sits ahead of price
Both spans are displaced twenty six periods to the right. That single detail causes most of the confusion around this method, because a shape drawn beyond the last candle looks like a prediction. It is not. The cloud sitting in front of price was computed from highs and lows that have already printed, and nothing about it will change as the next candles arrive. The cloud sitting behind price, over the candles you are reading right now, was computed twenty six periods ago.
So when somebody says price is above the cloud, the comparison being made is between today's price and a blend of ranges from well before today. Chikou Span does the same thing in reverse. Today's close is drawn twenty six periods back, so the familiar question of whether chikou is clear of price is really asking whether today's close is above the price that traded twenty six periods ago. That is a legitimate question. It is also entirely backward looking, and no amount of forward displacement changes that.
Five lines, one input
Here is the part worth internalising. Tenkan, Kijun, both spans and the thickness between them are all computed from the same series of highs and lows. They differ only in window length and horizontal shift. When every one of them points the same way, you have not collected five agreeing opinions. You have a single measurement of range, repeated at a few window lengths and then smeared across the time axis.
Charting is full of this mistake, and Ichimoku makes it easy because the plots look visually independent. The practical consequence is a weighting decision. If you keep any kind of confluence list, the whole stack should count as one entry on it, not five, and it should sit beside information that genuinely comes from elsewhere, such as where structure actually broke or what the dollar was doing while gold rallied. Counting correlated inputs as separate votes is the most common way a thin idea starts to feel like a strong one, and it is hard to notice from the inside because the chart looks busy and reassuring.
Flat levels and cloud thickness on gold
A flat Kijun appears when neither the highest high nor the lowest low in the window has changed for a while, which only happens when price has stayed inside an older range. Traders often describe these flat levels as magnets. The mechanism behind that is less mysterious than the language suggests. A flat midpoint marks the centre of a range the market has already spent real time inside, and ranges that have absorbed a lot of activity tend to get revisited. The level is interesting because of the range, not because a particular method happened to draw it.
Cloud thickness is simply the gap between the nine and twenty six blend and the fifty two period midpoint. The two disagree most after a sustained directional move, which is exactly when the cloud looks dramatic. A thick cloud therefore tells you that older ranges sat a long way from newer ones. It does not tell you that the area will hold. If you want to test that for yourself, mark the original range that produced a flat level on the live chart and watch how price behaves on the retests rather than trusting the shading.
Period counts meet the gold clock
This method was designed for a market with a defined daily session. Spot gold runs almost continuously for five days, pauses, and reopens. That changes what a fixed period count covers. Nine periods on an hourly gold chart is a slice of the week whose character depends entirely on where it falls, because an Asian stretch and a London to New York stretch do not produce comparable ranges.
On a daily gold chart there is a sharper problem. The candle boundary is set by the chart server time, and shifting that boundary rewrites every high and low inside the window, which rewrites all five plots. Two charts described the same way can show a visibly different cloud for no reason other than where midnight sits. The weekend adds one more wrinkle. A Sunday open away from the Friday close can establish a new window extreme without a single hour of trading between the two prices, and a fifty two period midpoint will carry that artefact for a long time after everyone has forgotten the gap.
Where the cloud misleads
Three failure modes recur. The first is lag by construction. A twenty six period midpoint cannot keep up with a fast repricing, so after gold reprices on a scheduled release the base line sits far behind and a pullback to it is an enormous pullback. The second is range behaviour. When price is oscillating, it crosses a flat cloud repeatedly and each crossing looks like a transition. The third is late confirmation. Chikou is drawn twenty six periods back, so any rule that waits for chikou to clear price is deliberately waiting a long time.
The cloud twist, where the two spans swap places in the forward section, is often read as a turning point. Remember that the twist is already fixed before price arrives, so it marks where old range data crosses, not where anything is about to happen. None of this makes the method useless. It makes it a context tool with a known delay, and a context tool with a known delay should never be the thing that puts you in a trade.
Keeping it as context, not as a trigger
A defensible use is narrow. Price holding above a cloud that is stepping upward tells you recent ranges have been migrating higher, which is a regime statement rather than an entry. A flat span marks a range centre you can test against price behaviour. Kijun gives you a slow reference that is useful for judging how stretched a move has become, in the same way any slow moving average does.
What the method will not give you is a reason to act at a specific price. If you want to use it, write down in advance what would make you wrong, and make that an actual level rather than a colour change. The honest caveat is the one from earlier in this article: because every plot derives from the same extremes, the apparent weight of evidence is an illusion created by line count. Strip the shading away, ask what the underlying ranges were doing, and you will have extracted everything the method actually contains without the feeling of certainty it tends to manufacture.
FAQ
Is a cloud crossover a buy or sell signal?
No. A crossover of the two spans is the moment a fifty two period midpoint and a blend of nine and twenty six period midpoints swap places, and because both are displaced forward it is already fixed before price reaches it. It describes how older ranges compare with newer ones. Treating that as an instruction asks far more of the arithmetic than the arithmetic can support.
Why are the default settings nine, twenty six and fifty two?
They come from a trading week that included Saturday, so twenty six was roughly a month of sessions and fifty two roughly two. Gold trades a five day week, so the numbers no longer map to anything calendar based. That is not a fatal problem, but it does mean the defaults carry no special significance and should not be defended as if they did.
Does Ichimoku work on fast gold timeframes?
The arithmetic is identical on any timeframe, so nothing breaks. What changes is usefulness. On fast charts the windows cover minutes, the lines react to single candles, and crossings multiply. Because the whole method is built from range extremes, it needs enough bars to describe a range, which is why most people who keep it apply it on hourly charts and above.
What does a thick cloud actually mean?
It means the twenty six period blend and the fifty two period midpoint are far apart, which happens after a sustained directional move has pulled newer ranges away from older ones. It is a measurement of past disagreement between two windows. It is not a measurement of how much support or resistance exists at that price.
Should I combine the cloud with an oscillator?
If you do, understand what you are adding. An oscillator built from recent closes and a cloud built from recent extremes still draw on the same price series, so the independence is partial at best. Something structurally different, such as how price behaved at a prior high or what a correlated market did, adds more than another line derived from the same candles.
ⓘ See these ideas on real price: open the free XAUUSD live chart.