Double Top, Double Bottom and Triple Top Patterns: How to Trade the Reversal

A double top is two peaks at about the same price with a dip between them, an M at the end of an uptrend, and it only turns bearish when price closes below the low of that dip (the neckline). A double bottom is the W-shaped mirror after a downtrend. A triple top adds a third failed peak at the same level. Below: the double bottom mirror, the triple top, a comparison with the head and shoulders, two worked examples with the arithmetic, and what the chart grader I build does with these shapes.

Quick Answer: What Are Double Tops, Double Bottoms and Triple Tops?

The table puts the double top and double bottom side by side. The triple top gets its own comparison further down, against the double top and the head and shoulders. All three sit with the other reversal and continuation shapes in the chart patterns library.

The last row is the one I'd tape to the monitor. Two equal highs are not a double top yet, and most of the time they never become one.

What Is a Double Top Pattern?

A double top forms when a stock in an uptrend rallies to a high, pulls back, rallies to about the same high again and fails a second time. Two peaks, one dip between them, an M. The low of that dip is the neckline. It's a bearish reversal pattern, and it needs a trend to reverse: StockCharts' ChartSchool page on the double top reversal asks for "a significant uptrend of several months" before it. Two equal highs inside a sideways range are just the top of the range.

The story is supply. Buyers pushed to the high twice and sellers met them there both times. The second failure is the warning. The neckline break is the proof, because that's where everyone who bought the dip between the peaks is wrong, and plenty of them have stops sitting just under it (the piece on stop-loss hunting explains why those stops cluster where they do).

Until that break you don't have a double top. You have two highs. Bulkowski's page on Eve & Eve double tops puts a number on it: "If you don't wait, there's a 60% chance that price will continue higher without confirming the double top." His double bottom pages give 48% for the mirror case. That's one researcher's sample, and it's still the best argument I know for waiting for the close.

Double tops show up on any timeframe, from a 5-minute chart to a weekly. Investopedia's double bottom page says the longer the gap between the two lows, the likelier the pattern is to work, and recommends daily or weekly charts for it. StockCharts says the longer a triple top develops, the more significant its break. So I'd give a daily or weekly M more weight than an intraday one. The neckline is a support level like any other, and once it breaks it often turns into resistance.

Anatomy of a double top: peaks, neckline, spacing and the 3% rule

• Trend
  An uptrend into the first peak. No trend, no reversal.
• Peak 1
  The first high. It sets the resistance the pattern tests again.
• The dip
  The pullback between the peaks. Its low is the neckline. StockCharts says the dip is generally 10% to 20% and wants at least 10%.
• Peak 2
  About the same price. StockCharts: "Usually, a peak within 3% of the previous high is adequate."
• Break
  A close below the neckline confirms it. Volume often expands here, which strengthens the read without being required.

On spacing, StockCharts says the time between peaks "can vary from a few weeks to many months, with the norm being one to three months," and that peaks too close together "could represent normal resistance rather than a lasting change." Bulkowski found most Eve & Eve double tops have peaks "in the 2 to 6 week range," then adds, dryly, "This is another guideline I ignore." For bottoms StockCharts prefers at least four weeks between the lows, and Wikipedia reads tops very close in time as consolidation.

Those numbers are written for patterns that take weeks or months to build, and a 10% dip on a 5-minute chart would be a crash. None of the sources I read gives an intraday rule, so this rough version is mine: two clearly separate swing highs, a pullback between them that's a real swing rather than a couple of candles, and enough time between the tests that price did something else in the middle.

The second peak doesn't have to match. Slightly lower is the textbook look: buyers couldn't get back to the old high. Slightly higher still counts if it stays within that 3% and price can't hold above the first peak. A close well above both that holds is a breakout, and the double top is dead. Bulkowski also sorts these by shape, Adam for a narrow, spiky turn and Eve for a wide, rounded one, and calls the Eve & Eve double bottom "what most chartists call the classic double bottom."

What Is a Double Bottom, and Is It Bullish or Bearish?

A double bottom is bullish. It's the mirror of the double top: after a downtrend, price drops to a low, bounces, drops to about the same low again and holds. Two lows, a high between them, a W. That high is the neckline, and the pattern confirms on a close above it. StockCharts counts a second low "within 3% of its predecessor" as valid and expects the bounce off the first low to be 10% to 20%. Investopedia allows 3% to 4%, and reads a slightly higher second low as a sign that the selling "came to an earlier end."

The psychology is the top in reverse. Sellers pushed to the same low twice and couldn't break it, and the traders who bought the first bounce, plus everyone who missed it, defend the level the second time. That's why the second low usually prints on less selling than the first.

Two things set bottoms apart from tops, and the sources agree on both. StockCharts says "Bottoms usually take longer to form than tops," and "Volume is more important for the Double Bottom Reversal than the double top." Its double bottom page wants clear evidence of volume and buying pressure accelerating on the advance off the second low. So on a W, I look harder at the volume panel than I would on an M.

Don't confuse it with a rounding bottom, a slow saucer with no sharp low. If you can point at two separate lows, it's a double bottom. One long rounded drift is a rounding bottom, and its trigger is the rim.

How reliable are double tops and double bottoms?

Reliability splits into two numbers: how often the shape confirms at all, and how often a confirmed one pays. The first is the non-confirmation figure from the double top section. For the second, Bulkowski's bull-market tables for stocks (read October 21, 2026) rank the four double top variants 10th to 19th of 36 patterns, with break-even failure rates of 20% to 25%. Break-even failure is his term for a confirmed pattern that doesn't move at least 5% in the breakout direction, roughly the cost of trading it. Bottoms do better. The Eve & Eve double bottom ranks 5th of 39 with 12% failure, and the spiky Adam & Adam version ranks 26th with 16%.

He also tracks how often price reaches the full measured move: 43% to 64% across the double top variants, 65% to 73% for the double bottoms. So treat the measured move as an estimate, not a minimum. On tops it's not far from a coin flip. And these are his idealized numbers. On the big average rises in his double bottom tables, he writes: "The numbers are based on hundreds of perfect trades, so your chance of duplicating such a rise on a consistent basis is probably zero."

You'll also see a "65% to 70% success rate" quoted around the web for these patterns. I couldn't find a study behind it.

On a double bottom, the most common failure is a neckline break on thin volume that drops straight back under the neckline within a few bars. That's the same event as a bull trap: a breakout that fills up with buyers at the worst price and then reverses through all of them. A neat W does nothing to prevent it, which is why the volume on the breakout tells you more than the symmetry of the two lows. Before you commit a stop to one, run it against this:

None of those rows is a probability. My rule of thumb: one reading from the failure column is a reason to trade smaller, and two or more means you're looking at a shape. A tweezer bottom at the second low, two candles wicking the same price, is a useful extra: the same rejection at candle resolution, pinning the level tighter than the swing lows do.

What volume should do, and what it doesn't have to do

The typical profile is the same on both patterns: heavier volume on the first peak or low, lighter on the second, expanding on the neckline break. Wikipedia's article on the double top describes it the same way, with the first peak on increased volume and the second attempt "on a lower volume." If you want a refresher on what those bars actually count, the guide to what volume means covers it.

Typical isn't required, though. StockCharts wants the double top's break to come "with increased volume and/or an accelerated descent," so a fast drop on ordinary volume counts. Bulkowski's confirmation is a close below the valley with no volume condition at all. And on Eve & Eve double tops he found the opposite of the usual rule: "Patterns with a rising volume trend show better performance." Heavier volume on a second peak doesn't cancel a double top by itself. Where volume does carry real weight is bottoms, for the StockCharts reason above.

How Do You Trade a Double Top?

The rules are short. Following them when the second peak is rolling over in front of you is the hard part.

1. Check the trend. An uptrend into the first peak, or there's nothing to reverse.
2. Mark the neckline at the low between the peaks, and check the peaks sit within about 3% of each other.
3. Wait for a close below the neckline. A wick through it isn't a close. If you get faked out a lot, StockCharts suggests stricter filters: "A price filter could require a 3% support break before validation. A time filter might require the support break to hold for three days before it is considered valid."
4. Pick an entry. The breakdown close, or a retest of the broken neckline from below. Bulkowski measured a pullback to the breakout price within 30 days on 64% to 65% of double tops, so roughly a third of the time it doesn't come within a month.
5. Put the invalidation stop above the highest peak. A new high above both peaks means the pattern was wrong. A tighter stop inside the pattern improves the ratio and gets hit more often. Whether that's worth it is your call.
6. Estimate the target. Subtract the pattern height (peak to neckline) from the neckline. Price often stops short of it.

Indicators can add to the read. Investopedia suggests checking for bearish RSI or MACD divergence at the second peak: price makes an equal high while the oscillator makes a lower one. The divergence explainer covers how to read it. It's supporting evidence and doesn't confirm anything on its own.

A worked double top, with the arithmetic

Made-up numbers on a daily chart. A stock trends up into $50.00 (peak 1), pulls back to $45.00, rallies to $49.60 (peak 2) and rolls over. The peaks are 0.8% apart, inside the 3% tolerance, and the dip is 10% deep, which passes StockCharts' at-least-10% check. The height is $50.00 minus $45.00, or $5.00, so the measured-move estimate is $45.00 minus $5.00, or $40.00. A daily candle closes at $44.70, below the neckline. That's the confirmation.

For sizing I'll use a common rule of thumb: risk 1% of a $25,000 account, or $250 a trade. Here are the two entries.

Entry A shows the drawback these patterns are known for. The stop and the target are both sized by the pattern height, so you risk about as much as you hope to make, which is why Investopedia calls the risk/reward ratio "a drawback" of them. Bulkowski's own adjustment (multiply the height by the hit rate before subtracting it) pulls the estimate up to between $41.80 and $42.85, depending on the variant, which makes A worse. Entry B fixes the ratio but not the odds: the retest doesn't always come, and a stop that close gets tagged more often. A measured move is a price target like any other, a level to manage the trade around, and the stop loss guide covers how to place the other end of it.

Trading the double bottom

Same steps, flipped. A downtrend into the first low, the neckline at the high between the lows, a close above it, a stop below the lower of the two lows (Investopedia puts it at the second low), and the height added to the neckline as the estimate. Investopedia's version enters "on a daily close above the first rebound's high." Two differences in practice. Give the volume on the break more weight than you would on a top. And expect a throwback to the neckline: Bulkowski measured one within 30 days on 65% (Eve & Eve) to 67% (Adam & Adam) of double bottoms.

Once the break turns into an uptrend, the chart is trending again, and that's where I find these patterns easiest to trade. If price builds a bull flag above the old neckline, that's a continuation setup on its own terms, with its own stop.

What Is a Triple Top Pattern?

A triple top is three peaks at roughly the same price, separated by two pullbacks, at the end of an uptrend. It's a bearish reversal pattern. StockCharts' page on the triple top reversal describes "three equal highs followed by a break below support," and like every reversal it needs "an existing trend to reverse." The highs don't have to match exactly. They "should be reasonably equivalent," and Wikipedia notes the peaks aren't always evenly spaced.

Every triple top starts life as a double top. StockCharts: "Before the third high forms, the pattern may look like a Double Top Reversal." What makes it triple is a double top that never broke its neckline. Price held above the low between the first two peaks, rallied a third time and failed at the same high again. Investopedia says the same thing from below: "Double bottoms may fail and become a triple bottom."

Which line confirms it is the one detail guides disagree on. StockCharts, Bulkowski and Wikipedia all use the lowest low between the peaks. Bulkowski's rule is that "The pattern becomes valid when price closes below the lowest valley in the pattern." Many other guides draw a neckline through both lows instead, which slopes when the lows differ and breaks earlier. I treat the lowest low as the confirmation and the sloped line as an early warning.

Until that close, a triple top is neutral. StockCharts points out that three equal highs also show up in an ascending triangle or a rectangle, and that the triple top "should be treated as a neutral pattern until a breakdown occurs." Lows that keep rising into flat highs make an ascending triangle, which leans bullish. If neither side wins, Investopedia says the pattern "will simply become a longer term range," a horizontal channel you trade differently.

Triple tops are rarer than double tops (Wikipedia and Investopedia both say so) and slow. As major reversals, StockCharts says they "usually form over a three to six month period," and "The longer the pattern develops, the more significant the ultimate break." Intraday versions exist too, just with smaller, faster swings. Bulkowski adds one I didn't expect: "More triple tops appear in bear markets than in bull markets."

The psychology is a longer version of the double top. Buyers took three runs at the same price and lost every time. Each dip buyer between the peaks has a stop under the lows, and by the third failure the longs who are left are tired. When support finally gives, those stops turn into sell orders at once.

Volume usually fades as it forms, with a twist. StockCharts: "As the Triple Top Reversal develops, overall volume levels usually decline. Volume sometimes increases near the highs." An expansion on the break "greatly reinforces the soundness of the pattern." Bulkowski's sample has volume trending down 62% of the time but "usually high beneath formation of each peak." One naming trap: a "triple top breakout" on a point and figure chart is a bullish signal and a different thing entirely. StockCharts flags the difference on its page.

What is a triple bottom?

The bullish mirror: three lows at about the same price after a downtrend, confirmed by a close above the highest high between them. Bulkowski's wording: "The pattern confirms as a true triple bottom once price closes above the highest peak between the valleys." Before the third low it looks like a double bottom, and three equal lows can also be a descending triangle or a rectangle. If the middle low is clearly lower than the other two, Bulkowski says "it could be a head-and-shoulders bottom," which the head and shoulders guide covers under the inverse version. If the last low sits above the second, he says to "expect better performance." The catch is scarcity: "Triples are rare, so price often continues down without confirming the triple bottom." Buying the third low before that close is the same bet as catching a falling knife.

Is a Triple Top Stronger Than a Double Top?

A lot of guides say yes, on the logic that three rejections beat two. Bulkowski's data says no, at least for tops. His triple tops page (statistics updated August 27, 2020, on 1,964 patterns) ranks the triple top 24th of 36, with a 25% break-even failure rate and a 14% average decline. He sums it up himself: "Triple tops are chart patterns with performance toward the bottom of the list. The failure rate is higher than I like to see, and the average decline is meager." Every double top variant ranks above it, none fails more often, and each falls a little further on average (the table below has the figures). A third failed peak doesn't make the top more reliable.

Bottoms are a different story. His triple bottom ranks 12th of 39 with 13% failure, between the Eve & Eve and Adam & Adam double bottoms. So "triple is stronger" is wrong for tops and roughly a wash for bottoms. His "average decline" runs from the day after the breakout to the ultimate low on perfect trades, so use these figures to compare patterns with each other, not to forecast your P&L.

The row I'd read first is the full-target hit rate. Fewer than half of Bulkowski's triple tops reached the measured move, which is the strongest case in the table for treating that target as an estimate and planning exits around it.

Double top vs triple top vs head and shoulders: telling them apart

The quickest test is the middle peak. Two peaks at about the same level is a double top. Three at about the same level is a triple top. Three where the middle one sticks up clearly above the other two is a head and shoulders. Bulkowski's line: "If the middle peak is above the other two, then you have a head-and-shoulders top." Investopedia makes the same cut, saying that in a triple top "the middle peak is nearly equal to the other peaks rather than being higher."

You'll sometimes read that a double top with a third, lower peak is "morphing into a head and shoulders." It isn't. A third peak near the same level makes a triple top, whether it's a touch lower or not, and Bulkowski says a last peak below the middle one is a reason to "expect a stronger decline." A head and shoulders needs the higher middle. On reliability, Bulkowski ranks the head-and-shoulders top 9th of 36 (19% failure, 51% reaching target), a step ahead of the best double top variant at 10th and well ahead of the triple top. That's the sourced version of "head and shoulders is more reliable."

How Do You Trade a Triple Top or Triple Bottom?

Same rules as the double top, with one extra trap: the third peak. When price is rolling over from a third failed high, it's tempting to short it right there, and a few guides list that as an aggressive entry. I don't take it, and most of the sources say wait. At the third peak the pattern is still neutral, and three equal highs are just as likely to be the top of a range or the flat side of an ascending triangle.

1. Check the trend. An uptrend into the first peak.
2. Mark support at the lowest low between the peaks. If the two lows differ, draw the sloped line through both as an early warning.
3. Wait for a close below support. Not the third peak, not a wick.
4. Pick an entry. The breakdown close, or a retest of broken support from below. Bulkowski measured a pullback within 30 days on 66% of triple tops and a throwback on 65% of triple bottoms.
5. Put the invalidation stop above the highest of the three peaks. Investopedia notes a stop placed inside the pattern improves the reward relative to the risk, at the cost of more small stop-outs.
6. Estimate the target. Height (highest peak minus support) projected down from support. Bulkowski's own method shrinks it: multiply the height by the hit rate, 49% for triple tops, before subtracting.

Breakdowns fail too. Investopedia's triple top page says "No pattern works all the time," and price can complete the pattern and then "recover and move above the resistance area." It suggests "a bearish MACD crossover following the third peak" as extra confirmation, and StockCharts' 3% or three-day filters work here too.

A worked triple top

Made-up numbers again, daily chart. A stock trends up into $80.00, pulls back to $72.00, rallies to $79.50, pulls back to $74.00 and rallies to $79.20. At $74.00 it was a double top that never broke its $72.00 neckline. The three peaks sit within 1% of each other, and the third is below the middle one, the case Bulkowski says points to a stronger decline. Support is the lowest low, $72.00. The height is $80.00 minus $72.00, or $8.00, so the full measured move points to $64.00. Bulkowski's version takes 49% of the height, $3.92, for a target of $68.08. The confirming candle closes at $71.40.

Entry A is the reward-to-risk drawback in plain numbers: risk $8.80 a share to make $7.40 if the full move comes, and $3.32 on the hit-rate target. That's why I'd rather wait for the retest, knowing it shows up on only about two out of three of Bulkowski's triple tops within 30 days and that a stop $1.20 away gets hit more often. A triple bottom runs the same arithmetic upside down, and its 74% hit rate is kinder: on an $8.00-tall triple bottom, his adjusted move is $5.92.

Common mistakes with double and triple tops

What Does SnapPChart Do With a Double or Triple Top?

I build SnapPChart, so this is the precise version, from the code. It doesn't detect double tops, triple tops or head and shoulders. It has no label for any of these three shapes, nothing measures whether two peaks are within 3% of each other, nothing scores a neckline, and it never computes a measured move. It does name some other shapes. Its scored signals include candlestick patterns such as a hammer, an engulfing candle or a shooting star, plus the cup and handle, and its pullback read can call out a bull or bear flag. It reads one chart screenshot: the trend structure, swing highs and lows, support and resistance, the EMAs, VWAP, MACD and whatever volume bars it can see.

What it grades is narrower than "reversal patterns," on purpose. It only grades with-trend momentum continuations. A long is a pullback in an established uptrend. A short is a rally in an established downtrend. It doesn't grade reversals as setups. Its direction rules never pick a short against an established uptrend, which takes the second or third peak off the table (a short-only profile still runs it as a short, and the counter-trend warning scores it down). By the same rule, a chart in an uptrend that just topped and is rolling over reads as a long that's stopping out.

Where it does engage is later, once the break has turned into a trend. When the chart shows an established downtrend (lower highs and lower lows, price under falling 9 and 20 EMAs and under VWAP), a 1 to 3 candle rally back into the broken neckline is a with-trend short it can grade. Right after the break it may still read the old uptrend, though its structure read can name the neckline break as a change of character at that price. That part is an observation with no trade call attached. A third or later rally inside the downtrend can be graded down by its "3rd Counter-Trend Retrace" signal, because by then the move is getting old. It's the same logic as the rally inside a dead cat bounce.

Bottoms are the mirror. No long at the second or third low while the downtrend holds (a long-only profile still runs it as a long, and the counter-trend warning scores it down). After the neckline break, once the chart shows an established uptrend, a 1 to 3 candle pullback toward the broken neckline is a long it can grade. The double bottom strategy page walks through that one: the neckline break, the uptrend that follows, the volume and the first pullback.

On a chart still going sideways under three equal highs, there's no trend to continue, and I won't promise it tells you "no trade" there. If its read finds no direction, it grades the chart as a long unless your profile is set to short-only. A long right under a level that's been rejected three times can be scored down by its rejection signals, and I wouldn't count on that happening every time. The pattern call stays yours.

It also sees only what's in the frame: no live data, no other timeframes, nothing scrolled off the left edge, and volume it can't read counts as missing data rather than weak volume. If the breakdown only makes sense with the whole M in view, put the whole M in the screenshot. The case for sticking to the trend is in the momentum trading strategy guide, the section on what AI can't see on a chart covers the limits of any screenshot read, and the AI chart analysis page covers what the read looks at.