Double top and double bottom: how to confirm an M or W pattern before you trust it

After the second peak prints, a double top still has one thing left to do: close below the low that sits between the two peaks. Until some candle does that, the chart shows two highs at roughly the same price. Charts get captioned "double top" at that stage all the time.

Most of what decides whether you call it confirmed is not in the pattern. It is in choices you make before you look: wick or body for the line, how close the peaks have to be, how far apart, and which timeframe's close you are waiting for. This piece goes through those one by one. The double bottom gets its own section, because a few things do not simply flip.

How do you confirm a double top? Close, not wick

Binance Academy's beginner guide to classical chart patterns (page marked "Updated Aug 12, 2026"; we checked it on 28 Sep 2026) defines it like this: "A double top forms when the price reaches a similar high twice without breaking higher on the second attempt." Then: "It's considered a bearish reversal pattern, typically confirmed when the price falls below the support level between the two peaks (the neckline)."

The Wikipedia entry on double tops and bottoms pins down where that line comes from: "The two peaks are separated by a minimum in price, a valley. The price level of this minimum is called the neck line of the formation." And it gives the same trigger: "The formation is completed and confirmed when the price falls below the neck line."

"Falls below" still leaves room for argument. Bulkowski, whose pattern statistics turn up in crypto tutorials (more on that further down), is stricter on his page about confirmed patterns. His worked example is a double bottom: "price has to close above the top of the peak between the two bottoms." Turn that upside down for a double top and you need a candle to close below the valley low.

Read that way, none of these counts yet:

• A wick pokes under the line and the candle closes back above it. ChartScout's double-bottom guide is blunt about the mirror case: "An intraday pierce does not count." Changelly's article on double tops in crypto makes the same point from the market side: "Crypto wicks can briefly break important levels before reversing. That's why confirmation matters."
• The candle is still open. Price below the line with forty minutes left on a four-hour candle tells you where price is right now. The close has not happened yet.
• Only a smaller timeframe has broken. That one needs its own section.

Which close? A 24/7 market has no closing bell

A stock has one close a day. Crypto trades around the clock, so "close" only means the end of a candle on whatever timeframe you picked. A one-hour candle can close under the neckline inside a four-hour candle that is still open and may yet finish above it. Both readings describe the same few hours.

So decide the timeframe before you go looking for the break, and state it when you say "confirmed": which chart, which candle, closed where. If you have not settled on a timeframe yet, start with which candlestick timeframe to use.

Where to draw the neckline: wick low or candle body

A double top has a single low between its peaks, so the neckline is a flat line through that one price. That is the most visible difference from a head and shoulders, where two lows are joined and the line often slopes.

The catch is which price you take for that low. Use the bottom of the wick and the line sits lower, so price has to fall further before it counts as broken. Use the bottom of the candle body and the line sits higher, and a "break" arrives sooner. Two people reading the same chart can end up one saying confirmed and the other saying not yet, and both drew a reasonable line.

When we sketch these we use the wick low, for a plain reason: price actually traded there. That is our habit. Nobody set it as a rule, and the body works too. The one approach that does not work is waiting until price sits between the two possible lines and then picking whichever suits you. One method, all the way across the chart.

One more thing that gets skipped: the low itself should be a proper pullback, a swing low you would pick out on the chart, not any short wick that happens to be handy. For how swing points are read, see higher highs and higher lows. And when someone posts a chart with the neckline already drawn, check which method they used before you look at their conclusion.

How close do the two peaks need to be, and how far apart?

Binance Academy says "a similar high". Wikipedia gives no percentage either. How similar is similar is your call.

The one concrete yardstick we found comes from Bulkowski's rules for a sub-type he calls the Adam & Adam double bottom: "The price variation between bottoms is small (the average is 1%, but allow variations)." And: "The twin valleys are usually several weeks apart (16 days is the median)." Those numbers were set on stocks, for one kind of double bottom. Whether they carry over to another market or another timeframe is yours to decide; no source settles it.

On spacing, Wikipedia has a more direct warning: "If the tops appear at the same level but are very close in time, then the probability is high that they are part of the consolidation and the trend will resume." A few neighbouring candles topping out at the same price is not something we would call an M.

What if the second peak is slightly higher than the first? The Academy definition says the price reaches a similar high "without breaking higher on the second attempt". Read strictly, a small overshoot is already a break higher. Read loosely, it is still "similar". Another call you have to make.

Before you mark anything up, write four things next to the chart: wick or body for the line, how close the peaks must be, the minimum number of candles between them, and which timeframe's close counts. Then use the same four on the next chart.

Double top neckline retest: price broke the line, then came back to touch it

After a close below the line, price sometimes swings back up to it. That move is usually called a retest or a pullback. The line used to be the support between the two peaks; what matters now is whether price can get back above it and stay there. For more on reading those levels, see finding support and resistance.

From there the chart tends to show one of two things:

1. Price touches the area and turns lower, with closes still under the line. The break stands and the description of the pattern still holds. How far it falls after that is not written anywhere on the chart.
2. Price closes back above the line and holds there. The earlier break now looks more like a false one, and the call should be withdrawn. How to read that case is covered in spotting a false breakout.

Which of the two happens more often in crypto, we cannot say. We have no data for that market, and a number without data is made up.

Double top price target: how the measured move is calculated

Many write-ups attach a target once the neckline breaks. Changelly describes the usual method: "Take the distance between the resistance level and the neckline. Then project that distance below the neckline."

A made-up example, only to show the arithmetic: both peaks near 100, the valley low at 90, so the height is 10. Project 10 below 90 and you get 80.

That 80 is the pattern's own height moved below the line. It says nothing about whether price goes there or stops there. Binance Academy's own caveat in the same guide: "No chart pattern guarantees a specific outcome; using them alongside other tools and sound risk management can support more informed trading decisions."

Failed double top: what an unconfirmed M turns into

On the same confirmation page, Bulkowski reports from his stock database: "For bull markets, 63% of double tops failed to confirm." That figure belongs to his sample and nowhere else. The only thing we take from it is that, in at least one carefully counted data set, unconfirmed double tops outnumbered confirmed ones. Whether crypto looks similar, nobody has shown us comparable data, so we will not guess.

An unconfirmed double top usually turns into one of these:

• Price never closes below the valley low and keeps moving between it and the two peaks. At that point it reads more like a range, with the peaks as the ceiling and the low as the floor.
• Price climbs through both peaks and closes at a new high. The two peaks become a pause in an uptrend, and the M was never there.

That is why, until it confirms, we would rather call it "two highs at about the same price". Once you name it an M, you start waiting for it to drop.

Double bottom breakout: does it need volume, and what else flips?

The Academy guide keeps this one short: "The double bottom is the mirror image of the double top. The price holds a similar low twice before moving higher."

Most of the above does turn upside down cleanly. The line runs through the peak between the two lows. You still choose wick high or body top and stick with it. You still decide how close and how far apart the lows need to be. And it still takes a close above the line, not a wick through it. A second low that holds above the first only says sellers did not push lower this time; the bottom is not confirmed until the line gives way.

Volume is where the sources part ways. For the top, Wikipedia gives the textbook sequence: "Price reaches the first peak on increased volume then falls down the valley with low volume. Another attempt on the rally up to the second peak should be on a lower volume." Its double bottom section says "Most of the rules that are associated with double top formation also apply to the double bottom pattern," then adds: "Volume should show a marked increase on the rally up while prices are flat at the second bottom." That last sentence carries a citation-needed tag in the article.

Does a double bottom need volume on the breakout? It depends on whose rulebook you use. ChartScout's crypto double-bottom guide makes it a hard condition: its system requires that "the breakout candle's volume must reach at least 105% of the recent average," and it quotes Schabacker: "the volume rule must be applied quite strictly in the case of a double bottom." Binance Academy's double-bottom definition leaves volume out. Bulkowski's confirmation rule looks only at a close above the peak between the two lows. So before you throw out a low-volume break, or trust one, check who set the volume threshold and which market it was set on.

It cuts the other way too. Bulkowski's 1% and 16-day figures were written for double bottoms. Using them on double tops is your decision.

Double top vs head and shoulders

A head and shoulders has an extra peak in the middle: three highs, the centre one tallest, two lows joined into a neckline. A double top has two highs, one low, one flat line. Drawing the line, waiting for a close, and reading the retest all work the same way, and the head and shoulders details are in head and shoulders: not a pattern until the neckline breaks.

Where do double top success rates come from?

Search either pattern and you will run into success rates, failure rates and the share of patterns that "reach the target". Follow the citations back and some of them end at Bulkowski. ChartScout's crypto double-bottom page, for instance, labels its failure and measured-move figures as Bulkowski data.

Those two figures match Bulkowski's own page on the Adam & Adam double bottom, which notes under them: "The above numbers are based on 1,154 perfect trades." That page does not say which market the trades came from. His confirmation page describes the database behind his research as "over 1,000 stocks covering the 1970s to 2011." So a page with crypto in the title can be quoting numbers that trace back to no crypto sample at all.

We do not publish a success rate for these patterns because we have no crypto sample to base one on. When you meet a percentage somewhere else, look for what market it was counted on, which years, and what rule decided that a pattern had formed or failed. If those answers are missing, read the number as the author's impression. Binance Academy's guide says much the same in one line: "No single chart pattern is reliable on its own. Patterns describe probabilities based on past behavior, not guaranteed outcomes."