Ichimoku Cloud: The Five Lines, the 9/26/52 Settings and Every TK Cross on Bitcoin

The Ichimoku cloud, or Ichimoku Kinko Hyo, is a set of five lines that shows trend direction, momentum and support or resistance on one chart. StockCharts credits it to journalist Goichi Hosoda, who published it in his 1969 book. Three of the lines are midpoints of the highest high and the lowest low over 9, 26 and 52 bars; two of them are shifted 26 bars into the future to draw the cloud, and the close is plotted 26 bars into the past.

How the Ichimoku cloud is calculated

Each line uses a midpoint, (highest high + lowest low) / 2, not an average of closes. That is the main difference from a moving average: a line stays flat as long as no new high or low enters its window.

The cloud, or kumo, is the area between the two Leading Spans. TradingView colors it green when Span A is above Span B and red when it is below. Because both spans are shifted forward, the cloud already extends 26 bars beyond the last price: it is drawn from data that is known today.

How to read the cloud

• Price above the cloud: the trend is read as up, and the cloud is the first support below.
• Price below the cloud: the trend is read as down, and the cloud is resistance above.
• Price inside the cloud: no clear trend; the thicker the cloud, the wider the zone of indecision.
• TK cross: the Conversion Line crossing the Base Line. StockCharts calls it “the classic signal” and notes that signals counter to the existing trend “are deemed weaker”, for example a bullish cross made below the cloud.

The daily chart of an S&P 500 ETF below shows the three positions over the 150 sessions to October 28, 2026: the close stayed below the cloud until April 10, 2026, closed above it from April 13, and closed inside it on only five sessions from late July to October 23, each time back above it within a few sessions.

These are reading rules, not tested results. StockCharts also notes that a TK cross “can also be rare in a strong trend”, so the pair can give few signals during the strongest moves.

Every TK cross on weekly Bitcoin since 2019

With the default 9/26/52 settings on weekly Bitcoin, the first full set of lines is available on February 11, 2019. From then to October 19, 2026 there were 17 crosses of the Conversion Line over the Base Line. For each one, we note where the close was against the cloud that week, then the change from that close to the close of the week of the next opposite cross, in the direction of the signal: a positive figure means price went the way the cross pointed.

Method, so the counts can be reproduced: weekly bars starting on Mondays, from August 21, 2017; the lines are computed on highs and lows exactly as in the formula box. A cross is a week whose Conversion Line ends on the other side of the Base Line from the last week where the two lines differed; weeks where they are equal, which is common with midpoints, are skipped and never count as a cross. The cloud of a week is the pair of spans computed 26 weeks earlier, and a close on or between its edges counts as inside.

Three things stand out. The four bearish crosses made above the cloud, the kind StockCharts calls weaker, were all followed by a rise before the next cross; on June 21, 2021 the price went from 34,700.34 to 54,659.00. The largest move, +258.47%, came after a bullish cross made inside the cloud in May 2020, which a strict reading would have skipped. And the last cross, bullish on August 31, 2026 below the cloud, was still open on October 19, 2026, with the price 0.23% below the cross close. Each case has 2 to 4 crosses on one market: this shows how the signals behaved, it does not measure an edge.

Settings: 9/26/52 or slower

StockCharts gives 9 for the Conversion Line, 26 for the Base Line and 52 for Leading Span B as the defaults, and says the settings “can be adjusted”. A slower set roughly doubles them. We ran one, 20/60/120 with a 30-week shift, on the same weekly Bitcoin series: it needs 150 weeks of history, so its first full value comes on June 29, 2020, and it gave only 4 crosses.

Fewer crosses and longer trades: the slower set stayed long through most of the 2020 to 2021 rise and short through most of 2022, but it also gave back its last signal, a bearish cross at 65,776.47 after which the price rose to 80,154.09. Four trades are an anecdote, not a reason to prefer one set.

Which timeframe?

The settings count bars, so the same rules work on any timeframe, with a different number of signals. On daily Bitcoin, from May 24, 2025 to October 28, 2026, the 9/26/52 lines crossed 25 times; the 6 bullish crosses made above the cloud were followed by a rise in 2 cases, with an average of −0.85% to the next cross. On the S&P 500 ETF, daily, from September 27, 2024, there were 17 crosses; the best one was the bullish cross of April 25, 2025, below the cloud, at 541.66, followed by a rise to 664.59 by the next cross on October 23, 2025 (+22.70%). Shorter timeframes give more crosses and more of them reverse quickly.

Does it work?

We found no published test that settles it. The tables above are one market and a few dozen crosses, without costs. Our bench measures chart patterns detected by rule against random entries, not indicators (see our methodology). If you use the cloud, the sounder use is as a trend filter, taking only signals in the direction of price against the cloud, combined with support and resistance levels; that is a common reading, not a tested one.

Other tools in the same family: the MACD, which also compares a fast and a slow line, and the parabolic SAR, a trailing stop that flips with the trend. More are listed with the trend indicators.