How to recognise them and what really happened after the breakout in 673 real BTC, ETH and SOL patterns since 2021.
53.6%
of the time price carried on in the direction of the pole after the breakout. Chance: 50%
49.3%the same breakout without a pole
673real patterns (±3.8)
What it is
A bull flag has two parts: a pole, a strong and fast rise, and the flag, a short pause in which price drifts down a little or sideways inside a narrow channel. The textbook reads it as a rest: when price breaks the channel upwards, the rise should continue.
A bear flag is the same thing upside down: a strong fall, a pause that drifts up a little and a breakout downwards. If the two lines of the pause converge instead of running parallel, it is called a pennant.
BTC/USDT, 1 hour, 6 August 2021. Pole, 9 hours of pause drifting down inside the channel, upward breakout and price kept rising.
How we detect it
To keep the statistics honest, the pattern is found with fixed rules and without looking at the future: at each candle only what was known at that moment is used. The rules, the same the app's tool uses:
Pole: a move of at least 5 times the average hourly volatility (ATR) within 3 to 12 candles.
Pause: 5 to 20 candles that retrace no more than half the pole and move, at most, at a quarter of its speed (a pause, not a reversal).
Channel: the lines of the highs and the lows of the pause. If they slope against the pole or run sideways, a flag; if they converge, a pennant.
Breakout: the first candle that closes outside the channel, on the side of the pole. That is where you decide.
We reviewed a sample by eye before giving any figure; the speed rule was added then, because some V-shaped reversals were slipping in as flags. Then the usual measure: two bands at ±1.5 ATR from the breakout close and which one price touches first within 48 hours, compared with chance in the same data.
What happened
"In favour": price touched the band in the direction of the pole first. The margin (±) is the 95% confidence interval. 1-hour BTC, ETH and SOL candles since 1-1-2021.
Pattern
Cases
In favour
Chance
Bull flag
257
53.3% ±6.1
49.6%
Bear flag
254
53.4% ±6.2
50.4%
Pennant (bull and bear)
162
54.3% ±7.7
49.9%
All
673
53.6% ±3.8
50.0%
The same channel breakout without a pole
16,615
49.3% ±0.8
50.0%
How to read it:
The pattern departs somewhat from chance: 53.6% against 50%. With 673 cases the margin still brushes 50%, but the direction repeats: bull and bear, flags and pennants, all between 53% and 54%.
What it adds is the pole. The same channel breakout, without the strong move before it, stays at chance (49.3%). Breaking a channel says nothing on its own; breaking it after a pole, something.
Something is not much: almost half the time price went the other way. An edge of 3 or 4 points only shows over many trades, with a stop in its place and a size that survives the streaks.
BTC/USDT, 1 hour, 29 September 2023. Clear pole, 13-hour flag and upward breakout… and price touched the lower band first. It happens in almost half the cases.
By coin and by year
BTC
ETH
SOL
In favour
51.7% ±6.3
51.7% ±6.4
58.2% ±6.9
Year
2021
2022
2023
2024
2025
2026
In favour
49.0%
55.5%
50.9%
53.8%
55.9%
56.2%
Five of six years above 50%, although each year on its own has few cases (90–137) and a margin of ±8 to ±10 points. SOL, the most volatile of the three, departs the most.
Test yourself
Three real flags or pennants picked at random, in their real proportion of hits and misses: we have not chosen the pretty ones. You see the chart up to the breakout candle and decide. Then, what happened.
Preparing the cases…
Historical data. Education, not financial advice. Getting it right here guarantees nothing in the real market.
Common mistakes
Seeing flags without a pole. A channel breaking without a strong move before it stayed at chance.
Mistaking a reversal for a pause. If the "flag" recovers more than half the pole, it is no longer a pause.
Entering before the breakout. The pattern is confirmed when a candle closes outside the channel, not before.
Believing a textbook pattern cannot fail. It failed in almost half the cases; the stop goes where the pattern stops making sense, on the other side of the flag.
Drawing conclusions from few cases. A single year has a margin of ±10 points: 49% in 2021 and 56% in 2026 fit within the same thing.
Check what you have learned
Five questions about this article. Each answer is marked straight away and explains why.
Knowing the theory is half of it. The other half is recognising the pattern with the chart in front of you and not knowing what comes next. Test yourself with 3 real cases or train it in Kronochart with patterns from every year.
Frequently asked questions
What is a bull flag?
A strong, fast rise (the pole) followed by a short pause inside a channel (the flag). The textbook expects the rise to continue when the channel breaks upwards.
Do flags work?
In 673 real patterns, price carried on in the direction of the pole 53.6% of the time (±3.8), against 50.0% for chance. Without a pole, the same breakout stayed at 49.3%.
Flag or pennant?
In a flag the lines run parallel or nearly so; in a pennant they converge. In the data they behaved similarly: 53.4% and 54.3%.
Where does the stop go on a flag?
Where the pattern stops making sense: on the other side of the flag, with some margin. How to size the position with that distance is in Risk management.
What about the "height of the pole" target?
Textbooks project the height of the pole from the breakout. Here we measure something more modest and comparable with the other studies: which ±1.5 ATR band price touched first. With more distant targets, the hit rate drops.
Train your eye with real patterns
In Kronochart you decide on real moments from the past, with flags, triangles and the other patterns in their real proportion of hits and misses, and then see what happened. With hundreds of decisions you will see whether your reading beats chance.