Head and Shoulders Pattern and Inverse Head and Shoulders: How to Trade Both

The head and shoulders is the first reversal pattern most traders learn, and the one I see called too early more than any other. Three peaks, a line under them, a close through that line. The inverse head and shoulders is the same shape flipped at the end of a downtrend. Below: both versions defined, what the neckline slope and the volume actually tell you, Thomas Bulkowski's hit rates, two worked examples with the arithmetic, and what the chart grader I build does with these charts.

Quick Answer: The Head and Shoulders and Its Inverse

Here are the two side by side, rules on top and Bulkowski's numbers underneath. Both sit with the other reversal and continuation shapes in the chart patterns library.

Two rows matter most. The confirmation row, because before that close neither shape is a pattern yet. And the full-target row: the top misses its measured move about half the time, the bottom about three times in ten. The ranks come from two separate lists (36 bearish patterns, 39 bullish ones), so don't compare them across the columns.

What Is the Head and Shoulders Pattern?

A head and shoulders is a bearish reversal pattern at the end of an uptrend. Price makes a high (the left shoulder) and pulls back, rallies to a higher high (the head) and pulls back to about the same level, then rallies to a lower high (the right shoulder) and rolls over. The two pullback lows are the armpits, and the line through them is the neckline. The pattern completes on a close below that line. StockCharts' ChartSchool page on the head and shoulders top is blunt about the precondition: "Without a prior uptrend to reverse, there cannot be a Head and Shoulders reversal pattern."

The size of that uptrend matters too. Bulkowski's rule on his head-and-shoulders tops page: "Price must have something to reverse, so if the rise leading to the pattern is small, expect a small decline." He also found that a short- to intermediate-term rise into the pattern gave the best results after the break. Three bumps after a modest two-week rise only have a modest rise to give back.

The story is buyers running out. The left shoulder is a normal high in a healthy uptrend. The head is one more push to a new high, usually on less volume than the left shoulder drew. Then the drop off the head takes price all the way back to the first armpit, the first real crack. The right shoulder is the tell: buyers try again and can't get near the head. Everyone who bought either armpit has a stop under the neckline, so when it gives, those stops turn into sell orders together. The neckline is a support level like any other, and once it breaks it often acts as resistance.

It shows up on any timeframe, from a 5-minute chart to a weekly, and how long it takes depends on the chart. Wikipedia's article says top formations are "completed in a few weeks," while a major bottom usually takes longer. None of the sources I read gives an intraday duration, so this rough rule is mine: three swings you could each point at on their own, with pullbacks between them that are real swings and not two candles of noise. I'd give a daily or weekly head and shoulders more weight than a 5-minute one, since it took longer to build and more traders are watching the same neckline.

Anatomy of a head and shoulders: shoulders, head, neckline and the break

• Trend
  An uptrend into the left shoulder. No trend, no reversal.
• Left shoulder
  A high in that uptrend, then a pullback. The pullback low is the first armpit.
• Head
  A higher high, then a drop back to about the first armpit. That low is the second armpit. The head has to stand clearly above both shoulders.
• Right shoulder
  A lower high than the head, often near the left shoulder's level. It doesn't have to match.
• Neckline
  The line through the two armpits. Flat, rising or falling.
• Break
  A close below the neckline confirms it. A wick through it isn't a close.

Symmetry is where the textbook and the data part ways. Bulkowski's identification guidelines say the shoulders should peak near the same price and sit about the same distance from the head, and StockCharts says "While symmetry is preferred, sometimes the shoulders can be out of whack." But his research, summarized on his head-and-shoulders explained page, points the other way for tops: "The more unsymmetrical a head-and-shoulders top appears, the better it performs." Tops with a higher left shoulder fell an average of 25% after the break, against 20% for a higher right shoulder and 19% for even shoulders. That's why the example above has a higher left shoulder. Symmetry helps you recognize a top. It isn't a quality filter. The finding flips for bottoms, which the inverse section covers.

Does it matter if the neckline slopes?

The neckline can be flat, rise or fall, depending on where the two armpits land, and guides disagree on what the slope means. StockCharts says "a downward slope is more bearish than an upward slope" on tops, and the reverse on bottoms. Bulkowski's neckline numbers back neither half. His tops with flat necklines did best, by a hair: average declines of 24% for flat, 23% for rising and 21% for falling necklines. On bottoms the falling neckline did best, 42% against 34% for the rest, the opposite of the StockCharts rule. I don't treat any slope rule as settled.

What the slope changes for sure is the trade. A rising neckline on a top gets crossed sooner and higher, so the stop above the right shoulder sits closer. A falling one gets crossed later and lower, so you give back more of the move before you're in and the stop sits farther away. Bulkowski handles the falling case with a stricter trigger: a top "confirms as a valid one when price closes below an up-sloping neckline or below the right armpit when the neckline slopes downward." Drawing the line is the same skill as any other trendline: two clear swing lows, extended to the right.

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

The typical profile is heaviest volume on the left shoulder, a little less on the head, lighter on the right shoulder, then an expansion on the break. Bulkowski found volume trending down across the pattern on 61% of tops. StockCharts says the left shoulder should outdraw the head "Ideally, but not always," and wants the break to come "in a convincing manner, with an expansion in volume." If you want a refresher on what those bars count, the guide to what volume means covers it.

Typical isn't required, though. About four tops in ten in his sample didn't have the textbook fade, and his confirmation rule is a close below the neckline with no volume condition at all. So on a top I read volume as supporting evidence. A heavy right shoulder makes me trade smaller. It doesn't cancel the pattern. Spot forex has no centralized volume, so on currency charts this whole section leans on tick volume at best.

What Is an Inverse Head and Shoulders Pattern?

An inverse head and shoulders is a bullish reversal pattern that forms after a downtrend: three troughs, the middle one (the head) deepest, with a neckline across the two reaction highs between them. It confirms on a close above that neckline. You'll also see it called a head and shoulders bottom (the name StockCharts and Bulkowski use), a reverse head and shoulders or an inverted head and shoulders. Same pattern.

The anatomy mirrors the top. A downtrend makes a low (the left shoulder) and bounces. Sellers push to a deeper low (the head), and price bounces again to about where the first bounce stalled. Those two bounce highs are the armpits. The right shoulder is a selloff that stops above the head, usually near the left shoulder's low, so it's a higher low. The neckline is resistance until price closes above it. Bulkowski's rule on his head-and-shoulders bottoms page mirrors the top: it confirms "when price closes above a down-sloping neckline or above the right armpit when the neckline slopes upward." So does his prior-trend rule: "if the decline leading to the pattern is small, expect a small rise."

The psychology runs the other way. The head is the sellers' last big push, and the bounce off it travels as far as the first bounce did. The right shoulder gives it away: sellers try again on lighter volume and can't get back to the low. Anyone short from higher up has a stop above the neckline, so a close through it brings their buy-to-cover orders in alongside the new buyers.

Two differences from the top show up in the sources. Bottoms tend to take longer to form: Wikipedia says a major bottom "may be prolonged for a period of several months or sometimes even more than a year." And the symmetry finding flips. Bulkowski found that "The more symmetrical a head-and-shoulders bottom appears, the better it performs," measuring how evenly the shoulders sit either side of the head. Don't mistake it for a triple bottom, three lows at about the same level, which the double top, double bottom and triple top guide covers. An inverse needs a head clearly below both shoulders.

Volume on an inverse head and shoulders

Volume carries more weight on a bottom, and here the sources agree. StockCharts' page on the head and shoulders bottom puts it plainly: "While an increase in volume on the neckline breakout for a Head and Shoulders Top is welcomed, it is absolutely required for a bottom." Its profile: heavy volume on the left shoulder's decline (it can stay heavy into the head), an increase on the advance off the head, light volume on the right shoulder's decline, and an expansion on the advance from the right shoulder and on the breakout. Bulkowski's bottoms have volume trending down across the pattern 65% of the time, highest on the left shoulder or head and diminished on the right shoulder.

So the volume test I apply is lopsided. On a top, a quiet break makes me smaller. On an inverse, a quiet break makes me pass. Relative volume is the easiest way to tell a real expansion from an ordinary bar.

How Reliable Is the Head and Shoulders Pattern?

Reliability splits into two questions: how often a confirmed pattern fails, and how often it reaches its measured move. The only sourced numbers I found are Bulkowski's bull-market statistics on stocks. On his head-and-shoulders tops page (statistics updated August 26, 2020), the head-and-shoulders top ranks 9th of 36, has a 19% break-even failure rate and reaches the full measured move 51% of the time, on more than 2,800 trades. 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. His head-and-shoulders bottoms page ranks the head-and-shoulders bottom 13th of 39, with 11% failure and 71% reaching the target, on 3,197 trades. His summary: "Head-and-shoulders bottoms are reliable chart patterns that sport a low failure rate and good average rise."

So is the inverse more reliable? In his data, yes, on the two numbers that compare cleanly: bottoms failed less often (11% against 19%) and reached the full target more often (71% against 51%). Skip the ranks for this, since they come from separate lists. The average moves don't compare either. A 45% rise and a 16% decline don't sit on the same scale, since a decline can never pass 100% and a rise has no ceiling.

Two caveats before you lean on any of it. These are what he calls perfect trades, measured from the day after the breakout to the ultimate high or low, which nobody trades in practice. And they're stocks only, so they say nothing about forex or crypto. Use them to compare patterns with each other, not to forecast your P&L. You'll see much higher success rates quoted around the web. I couldn't find a study behind any of them.

Retests are common, and they cut both ways. Bulkowski measured a pullback to the breakout price within 30 days on 68% of tops and a throwback on 65% of bottoms, and both pages carry the same warning: "Pullbacks hurt post breakout performance" on tops, "Throwbacks hurt postbreakout performance" on bottoms. A retest is a cheaper entry. It isn't extra confirmation.

How Do You Trade a Head and Shoulders?

The rules fit on a sticky note. Sticking to them while the right shoulder rolls over in front of you is the hard part.

1. Check the trend. An uptrend into the left shoulder, big enough to have something to give back.
2. Mark the neckline through the two armpits, and check the head clears both shoulders.
3. Wait for a close below the neckline. A wick isn't a close. If the neckline slopes down, Bulkowski waits for a close below the right armpit instead. His one-line tip: "Wait for confirmation before placing a trade."
4. Pick an entry. The breakdown close, or a retest of the broken neckline from below. Some traders short the right shoulder as it rolls over, with a tight stop just above it. That's the aggressive entry, and it's a forecast: the pattern hasn't confirmed, and plenty of right shoulders turn back up before the neckline while the uptrend carries on.
5. Place the stop. Above the right shoulder is the usual invalidation, since a new high there erases the lower high the pattern is built on. Above the head is the wider stop, safer from noise and much worse on reward to risk. A stop just above the neckline is the tight option. It's cheap, but retests land right on that line, so it gets hit a lot. Each one is a trade-off you choose on purpose.
6. Estimate the target. Head to neckline, subtracted from the breakout price. It's an estimate, not a minimum. StockCharts: "Any price target should serve as a rough guide." Bulkowski's own measure rule multiplies the height by his hit rate (51% for tops) before subtracting it.
7. Plan the exit before the target. With a target that's reached about half the time, I'd cover part of the position at the hit-rate estimate, move the stop on the rest to breakeven, and trail it above each new lower high. The guide to scaling out of partial profits and the trailing stop guide cover both halves of that.

Indicators can add weight without confirming anything. Bearish RSI or MACD divergence at the head (price makes a higher high, the oscillator a lower one) is the classic one, and the divergence explainer covers how to read it. A MACD cross down on the break and price under falling short EMAs fit the same story.

A worked head and shoulders top, with the arithmetic

Made-up round numbers on a daily chart, the same ones in the first diagram. A stock trends up into a $91.00 left shoulder, pulls back to $84.00, rallies to a $96.00 head, drops back to $84.00 and puts in a $89.50 right shoulder. The neckline is flat at $84.00. The height is $96.00 minus $84.00, or $12.00, so the full measured move points to $72.00. Bulkowski's version takes 51% of the height, $6.12, for an estimate of $77.88. A daily candle closes at $83.40. 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. The position sizing guide explains the shares row. Three ways in:

Column A is the drawback these patterns are known for. The stop and the target are both sized by the pattern, so on the hit-rate estimate you risk $6.30 to make $5.52. Moving the stop above the head (A2) doubles the risk for the same reward. B fixes the ratio but not the odds: a retest shows up on about two-thirds of tops, and a stop $2.00 away gets tagged more often. That ratio is why I'd only take the break if the hit-rate number still clears my minimum reward to risk. Here it doesn't.

Trading the inverse head and shoulders, with an example

Same steps, flipped. A downtrend into the left shoulder, the neckline through the two reaction highs, a close above it, then an entry on the breakout close or on a throwback to the neckline from above. The usual stop goes below the right shoulder's low, since a new low there erases the higher low the pattern is built on. Below the head is the wider option. The target is the head-to-neckline height added to the breakout price, an estimate Bulkowski's bottoms reached 71% of the time. Give the breakout volume more weight than you would on a top, for the StockCharts reason above.

Made-up numbers again, daily chart, matching the second diagram. A stock falls into a $30.00 left shoulder, bounces to $34.00, drops to a $26.00 head, bounces back to $34.00 and holds a $30.00 right shoulder. The neckline is flat at $34.00 and the height is $8.00, so the full measured move is $42.00, and Bulkowski's version (71% of the height) is $39.68. A candle closes at $34.50 on expanding volume.

On the hit-rate rows the bottom's ratios beat the top's (1.10, 0.60 and 3.79 against 0.88, 0.43 and 3.06), because the higher hit rate keeps more of the measured move in play. On the full-estimate rows the top's are slightly better (1.81, 0.89 and 6.00 against 1.60, 0.86 and 5.33). Entry A risks $4.70 to make $5.18 on the hit-rate estimate. The stop below the head nearly doubles the risk and pushes the ratio under 1. The throwback entry is the cheapest by far if the throwback comes, which it did on 65% of Bulkowski's bottoms, and if you accept more stop-outs with $1.50 of room.

What a failed head and shoulders looks like

Breaks fail. On a top, the first warning is a close back above the neckline after the break, and the pattern is plainly wrong once price clears the right shoulder, which is exactly why the stop lives there. On an inverse it's the mirror: a close back below the neckline, then a break of the right-shoulder low. For bottoms, StockCharts adds a volume test: "Without the proper expansion of volume, the validity of any breakout becomes suspect." And one of Bulkowski's trading lessons on his bottoms page is titled, simply, "Tall one-day breakout leads to disaster."

A break that reverses straight back through the neckline is a bull trap or bear trap, depending on the side, and it fills up with traders at the worst price. A failed pattern doesn't promise a big move the other way, though. Some guides call it a powerful signal in its own right. I haven't seen data for that, so I read a failed top as an uptrend that hasn't ended yet and judge any trade from there on its own evidence.

Is a Head and Shoulders Stronger Than a Double Top?

Tell them apart by 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 both shoulders is a head and shoulders, or in Bulkowski's words, "A three-peak pattern with the middle peak above the others." So a double top that develops a third, lower peak hasn't turned into a head and shoulders. Neither of its first two peaks stands above the other, so there's no head. A third peak near the same level makes a triple top, and a clearly lower one is just a lower high after the double top.

On strength, Bulkowski's numbers give the head and shoulders a small edge, not a different league. His top ranks 9th of 36 with 19% break-even failure. The best of his four double top variants, the Adam & Eve double top, ranks 10th with 21%. The double top guide shows the range across his four double top variants next to the head and shoulders and the triple top. As for the idea that the extra shoulder adds confirmation: the right shoulder does show buyers failing below the high, but nothing confirms either pattern until the close through the neckline.

Two more lookalikes. Bulkowski warns that a top "could really be a broadening formation, right-angled and ascending," a flat bottom under rising highs. And a pattern with extra shoulders on one or both sides is a variant he files separately as the complex head and shoulders.

Common mistakes with head and shoulders patterns

What Does SnapPChart Do With a Head and Shoulders?

I build SnapPChart, so this is the precise version, from the code. It has no head and shoulders classifier. It doesn't label, detect, draw or score the pattern, and nothing in it measures the neckline, its slope, the symmetry of the shoulders or the volume across them, or computes a measured move. 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. Its scored signals include some named candlestick shapes, such as a hammer, an engulfing candle or a shooting star. Separately, its structure read describes how the last few candles reacted at the key level. That description is shown to you and doesn't set the grade.

What it grades is narrower than "reversal patterns," on purpose: with-trend momentum continuations only. A long is a pullback in an established uptrend. A short is a rally in an established downtrend. It reads every chart by its established trend, so while the uptrend is still intact, it won't pick a short into the right shoulder. It reads the chart as a long, and a top that's rolling over inside an uptrend as a long that's stopping out, not a short. If your profile is set to short only, it runs the chart as a short anyway, and its counter-trend warning scores that short down.

Where it does engage is after the break. Once the neckline has broken and the chart shows lower highs and lower lows, with price under falling 9 and 20 EMAs and under VWAP, a 1 to 3 candle rally back toward the broken neckline is the kind of with-trend short it can grade, with an entry, a stop and a target. Right after the break it may still read the old uptrend, though its structure read can describe the neckline break as a change of character at that price. That's an observation with no trade call attached. A third or later rally inside the new downtrend can be scored down by its "3rd Counter-Trend Retrace" signal, because by then the move is getting old.

The inverse is the mirror. No long at the right shoulder while the downtrend holds (a long-only profile still runs it as a long, and the counter-trend warning scores it down). After a close above the neckline, once the chart shows higher highs and higher lows, a pullback toward the broken neckline is a long it can grade. The page on the confirmed head and shoulders setup walks through what it reads on that kind of chart, for both directions.

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 trend change only makes sense with all three peaks in view, keep them 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.