Патерн голова і плечі: як знаходити і торгувати

Everyone draws this pattern. Open any educational channel and there they are — three bumps, a line underneath, and an arrow pointing down labeled "profit." In real trading the head and shoulders pattern fails roughly a third of the time, and it fails not because the pattern is bad, but because the trader took the entry off a picture without checking whether there was any money behind it.

Head and shoulders doesn't work like a magic signal. It works like a map of stacked stops. Below the neckline sit the protective orders of everyone who bought the last impulse, and the market sees that perfectly well. The only question left is whether there's enough aggression to reach those stops.

Let's take the pattern apart. How it forms, where to actually put the entry and the stop, how to tell a real breakout from a painted one using the order book and the tape, and why the "measured move" target hits far less often than advertised.

What the Head and Shoulders Pattern Is

The head and shoulders pattern is a reversal formation made of three consecutive extremes, where the middle one is higher or lower than the two outer ones. The middle extreme is called the head, the side ones are the shoulders, and the line connecting the intermediate lows (or highs in the inverted version) is the neckline.

Richard Schabacker described the pattern back in the 1930s, and it became common knowledge after the 1948 Edwards and Magee book. For almost a hundred years the same construction has been passed from textbook to textbook. That's exactly why it works. The whole market sees it.

Keep your selection criteria strict. The formation has to be logical, obvious, and clear not only to you. Ask yourself whether another trader would see the same picture on the same chart. If the answer is "probably not," you drew it in yourself.

The mechanics are simple. While the trend is up, every next high is above the previous one. The head still respects that order. The right shoulder doesn't. It sits below the head, which means buyers failed to push price further twice in a row. Then the neckline break produces a lower low, and the structure of the up move is officially broken.

Stops. That's the main thing. The trader who shorted from the head put a stop above the high. The trader who went long from the neckline put a stop below it. The pattern here is direct: most people learned from the same material and place protective orders in similar spots. A cluster of stops is guaranteed liquidity, and price gets pulled toward liquidity.

Classic (Reversal at the Top)

A reversal structure after a rally. Three peaks, the middle one higher than the sides, the neckline drawn through the two lows between them.

Let me use numbers so it isn't abstract. BTC/USDT, 15-minute chart.

The left shoulder forms at 68,400, pullback to 67,500. Then an impulse to 69,200 — that's the head — and a pullback again to 67,480. The neckline comes out almost horizontal, around 67,490. The third attempt to rally runs out of air at 68,350, slightly below the left shoulder. The right shoulder is done.

The height of the head above the neckline is 69,200 − 67,490 = 1,710 points. That's the base for the target calculation, we'll come back to it.

The proportions I look at first. The right shoulder should come in at roughly 60-90% of the head's height (measured from the neckline). If it nearly reaches the head, the pattern is loose and the seller isn't as strong as it looks. If it's very low, the move already started without you, and a breakout entry gives a terrible risk-to-reward.

In time, the shoulders are ideally symmetrical. In practice in crypto the right shoulder often forms faster than the left, especially if the reversal comes after a news spike. Asymmetry of one and a half to two times is normal; five times is a reason to be careful.

One more detail almost nobody writes about. A good classic head and shoulders pattern almost always has a head that pushed past a round number or a previous significant high. Same logic. The impulse reached the zone where large limit orders for profit-taking were sitting, and that's where it got killed. At the very top of the head you'll often see a candle with a long upper wick and a small body — there's a separate breakdown of those signals in the article on candlestick patterns.

Inverted (Reversal at the Bottom)

The mirror structure after a decline. Three lows, the middle one below the sides, the neckline drawn through the highs between the troughs. It's also called the inverse pattern.

There's a fundamental difference from the classic version, and it concerns volume. In a top reversal, volume usually fades from the left shoulder to the right one — the seller doesn't need much effort for price to slide. In a bottom reversal, you need rising volume off the right shoulder. Without it, a neckline break to the upside is almost always false and price drops right back.

The reason is physical. A decline can happen on its own, under its own weight, but a rally takes money. Somebody has to actually buy.

In crypto, the inverted pattern has an extra accelerator: a cascade of short liquidations. A liquidation is a forced position close by the exchange when the loss has eaten the margin. While price grinds out the second bottom, shorts pile in with leverage, and their stops and liquidation levels collect above the neckline. The break triggers the forced closes, and the move comes out sharper than calculated. Hence the practical consequence: the inverted pattern overshoots its target more often than the classic one.

The liquidation map confirms this directly. If there's a dense cluster of potential short liquidations hanging above the neckline, the breakout has fuel. If it's empty, you're relying on organic demand alone. We covered the mechanics of forced closes separately, in the article on futures liquidation and the liquidation map.

How to Find It on the Chart

First selection rule. Below the 15-minute chart, looking for this pattern is pointless. On the 1-minute chart a "head and shoulders" gets drawn ten times an hour out of random noise, and none of them collect enough stops for the move to cover fees plus the spread.

The working timeframes for crypto are 15 minutes, 1 hour, and 4 hours. On the 5-minute chart the pattern is acceptable only on very liquid pairs and only as part of a larger structure.

Second rule — the coin. The numeric criteria for an actively tradable instrument: daily volume above $100-150 million, more than 800,000 trades, daily price change of 10-15% or more. On a coin turning over five million you can find a geometrically perfect pattern, but you won't fill it at your calculated prices — the order book is empty.

Here's the checklist I use to throw out the garbage before I even start thinking about a trade.

• There's a clear trend ahead of the pattern, at least two or three impulses. There's nothing to reverse if the market was already chopping sideways
• The head is noticeably above both shoulders, not 0.3% above. A difference smaller than half the ATR isn't a head, it's three identical touches of a level
• The neckline has at least two touches, and both are visible without zooming in
• The right shoulder didn't take out the head's high with any wick
• The formation took at least 25-30 bars. A seven-candle pattern hasn't collected anyone
• The coin has adequate turnover and a live tape

Six points cut the number of candidates radically. Out of a dozen "pretty pictures" per session, one is left, two at most. That's normal.

The Neckline

The neckline is the most important part of the structure and at the same time the most sloppily drawn. Its break is what counts as the signal, so both the entry point and the stop depend on how precisely you draw it.

It's built from the two intermediate extremes. For the classic pattern, from the lows between the shoulders and the head; for the inverted one, from the highs. It's more correct to run the line through candle bodies, not wick tips. A wick is an outlier, a random touch, while the body shows where the market actually traded.

The terminal has a dedicated toolset for this. Line [F] for horizontal levels, Trend Line [T] for sloped ones, Ray [R] when you need to extend a level to the right indefinitely, Rectangle [E] to mark a zone instead of a thin line. Plus the Magnet (LeftCtrl), which snaps the anchor point exactly to a candle's high or low and removes the shake in your hand. A small thing, but when the stop is counted in ticks, that small thing costs money.

A separate topic is what counts as a break. Touching the line isn't a break. A wick poking through isn't a break. The working criterion is this: a bar closes beyond the line by at least 0.2-0.3 ATR of the current timeframe. Anything less is noise inside the level.

A retest after the break happens roughly half the time. Price drops, then comes back up to the neckline from below, pokes at it, and continues. Former support acts as resistance. There's a detailed breakdown of how a level flips roles in the article on support and resistance levels.

I usually split the position. Half goes in on the break, half is held in reserve for the retest. If the retest doesn't happen, I work with half size and don't lose sleep. If it does, I add at a better price and set the stop for the whole position beyond the retest level.

Volume During Formation

Volume is the filter that separates a real pattern from a painted one. It works differently for the two versions of the pattern, and this is where most people get confused.

Divergence between price and volume at the head is the most valuable signal in the whole structure. Price made a new high, but volume is lower than on the previous impulse. That means there's no money behind the new high — it was pushed there on leftover momentum.

In crypto, volume alone isn't enough, because it doesn't show who was trading. That's where cluster charts come in. A cluster is the same candle, but with volume broken down by price levels inside it, plus delta. It answers the question of where liquidity actually changed hands and who controlled the level.

What to watch in the clusters as the pattern forms:

• At the head. A lot of volume but price barely moved — that's absorption. Buyers are hitting with market orders while someone is taking all of it with limit orders and refusing to let price go higher
• At the right shoulder. Delta should skew toward the seller before the break, not at the moment of the break. If the skew only shows up on the break itself, you're entering a move that's already played out
• At the neckline on approach. If cluster volume at that level drops sharply, the buyer is gone and the level is empty. The break will be easy
• After the break. Large volume on the first bars below the line confirms the stops actually filled

The cluster chart solves one more problem. It shows whether an order in the book was real. A large density level sat in the book under the neckline, price came in, real volume printed at that level in the cluster — the order was live and it got eaten. The order disappeared before price arrived and there's no volume in the cluster — that was spoofing: a fake order posted to scare people off and pulled before it could fill.

If the order book is still a dark forest for you, watch the free lesson on limit orders, liquidity, and entry points. It's part of our course "Trading from Scratch | free crypto trading and scalping course" on YouTube — all five lessons are open, no payment and no registration.

How to Trade It

First, the unpleasant truth about the statistics. The pattern doesn't play out 80% of the time the way the courses promise. In my experience in crypto, the pattern fails roughly a third of the time, and the lion's share of those failures comes from setups that traded against the higher timeframe trend.

Hence the basic filter before any entry. Check the direction on the timeframe four times higher than your working one. Trading the pattern on 15 minutes — open the hourly. If the higher timeframe is pointing up and you're planning to short a classic pattern, the odds drop sharply. You can work against the higher timeframe direction, but in those cases I cut size in half.

Entry Point, Stop, Target

Three entry options, each with its own price and its own risk.

The stop. The textbook says to put it above the right shoulder's high. For scalping that's often far too far — the distance comes out at 1.5-2% and kills the whole economics of the trade.

Here's the working logic. Calculate the base distance from ATR (1.5x multiplier for intraday), which gives you a calculated point. Then look at the order book above that point and find the nearest density level. A density level is a cluster of limit orders at one price, noticeably larger than the average order size for that coin. Put the stop beyond it with a few ticks of cushion. Bare math gives you the distance; the order book gives you the specific level where somebody is there to protect you.

If there are no orders for another two ATR beyond the calculated point and the order book is empty, it's better to skip the trade. A stop in a vacuum fills with several times more slippage than you expect.

The target. The classic measured move formula.

Target = Neckline break point − (Height of the head above the neckline)

for the inverted pattern:

Target = Break point + (Depth of the head below the neckline)

The formula gives you a reference, not a guarantee. It fills completely about half the time on the classic pattern and more often on the inverted one (short liquidations help there). So don't just plot the target — check what's standing in the way.

The check goes like this. Plot the calculated target. Open the density map and see whether there are large limit orders between the entry and the target. If there's a $10 million density level sitting on the way and the average cluster volume on that coin is $4 million, price most likely won't reach the target. Take profit in front of the density level.

Case study. Classic pattern on BTC/USDT, 15-minute chart.

Same data we went through above. Head at 69,200, neckline at 67,490, right shoulder at 68,350. ATR(14) on 15 minutes is 240.

Price pulls back from the right shoulder and approaches the neckline. I check the order book. Below 67,490 there are almost no orders for seven ticks; the nearest notable density level is at 67,300. The tape accelerates, sells coming through at 3-8 BTC.

The bar closes at 67,420 — 70 points below the line, about 0.29 ATR. The break criterion is met.

Entry at 67,415 with a market order. Calculated stop from ATR × 1.5 = 360 points, which puts it at 67,775. I look higher in the book, find a density level at 67,810, and place the stop at 67,830. Final distance: 415 points.

Account $10,000, risk 0.5% ($50). Size = 50 / 415 = 0.12 BTC.

Target by the formula: 67,490 − 1,710 = 65,780. I check the density map — there's an 18 BTC order sitting at 66,200 that's been there for over forty minutes. First take goes at 66,250, in front of it. Distance 1,165 points, risk-to-reward 1 to 2.8.

How it played out. Price reached 66,280, bounced off the density level, pulled back to 66,900. I took half at 66,250 and moved the stop on the rest to breakeven. Then the density level got pulled, price went to 65,810, and I closed the second half there. Total around +$155 on $50 of risk.

If I had just parked the take at the calculated 65,780 and never looked at the order book, the position would have shown +2.5% first, then given it all back, and I'd have exited at breakeven. The difference came down to one look at the density map.

Confirmation Through the Order Book

The chart shows what already happened. The order book shows what's sitting there right now. The difference matters, and it's exactly what separates an entry off a picture from an entry off the facts.

Three tools answer three different questions. The order book tells you where the liquidity is. The tape tells you who's more aggressive right now. The cluster chart confirms where volume actually traded and who won. Together they turn geometry into a trading decision.

Here's what it looks like as price approaches the neckline.

Situation 1. A large density level sits below the neckline. The break may not happen at all. Somebody is holding the level with limit orders. If the density level is $12 million and average cluster volume is $5 million, one move isn't going to get through it. The scenario flips here. Instead of shorting the break, you can look at a bounce up off the density level, with the same requirement for confirmation: the buyer showing up on the tape as price approaches.

Situation 2. It's empty below the neckline. The order book is empty, no orders for several ticks. That's exactly the fast-break scenario — price will fall through with no resistance. Good for the entry, bad for the stop fill, so calculate size from the real distance, not the one you'd like.

Situation 3. A density level appeared and immediately vanished. Classic spoofing. The order was shown to scare people and pulled before price got there. You check it on the cluster chart. If no volume traded at that level, the order was fake.

What the tape should show at the moment of the break:

• Flow acceleration. Not isolated prints, but a rising frequency of trades
• A skew toward the seller by volume, not by count. A hundred small sells are weaker than two large ones
• No opposing aggression. If large buys show up on the tape during the break, someone is absorbing the volume and the break may turn out false

The Secret Terminal tools that cover this work.

• Density map. Shows limit orders that have been sitting in the book for more than 30 minutes, across all coins at once. A ready-made list of levels for your stop and take, no need to click through order books by hand
• Ruler (R). Measures distance in points or percent right inside the order book window. Measure the head height on the chart, carry it over to the book, and see how many levels the calculated target covers and which orders fall inside it
• Auto-stop. Places the stop at a set distance as soon as the entry fills. On a neckline break your hands are busy, and this helps noticeably
• P, L, O filters with the mouse wheel. Adjust the significance thresholds for orders and clusters. When volatility doubles, the old thresholds turn the order book into a string of Christmas lights
• Key C. Auto-tunes the order book preset to the coin's current volatility
• Chart trading. Stop and take are dragged with the mouse right along the neckline, without switching to the order book. Key G moves the stop to breakeven, J closes the position from the chart window

Case study. Inverted pattern on an altcoin.

A top-50 coin, $180 million daily turnover. Hourly chart, third day of decline.

Left shoulder at 1.842, bounce to 1.905. Head at 1.798, bounce to 1.912. Right shoulder at 1.851. The neckline slopes up, from 1.905 to 1.912, which by itself weakens the signal a bit.

The depth of the head below the neckline is about 0.112. So the calculated target after the break is somewhere around 2.025.

What the order book looked like. Above the neckline, at 1.925 and 1.940, two sell-side density levels were hanging, both worth several million, both sitting there for over half an hour. Formally that's an argument against the long. But the liquidation map showed a dense cluster of short liquidations in the 1.93-1.96 zone, and the volume there was higher than in the limit orders.

I waited for buyers to start eating through the first density level. The tape went green — about $1.4 million of aggressive buying came through in two minutes and the order at 1.925 dissolved. Entry at 1.9285.

Stop below 1.906, beyond the neckline and beyond the nearest buy-side density level. Distance 0.0225, or 1.17%. $50 of risk gives a size of 2,222 coins.

What happened next. The density level at 1.940 got pulled by its own owners (they clearly didn't want to meet the cascade), price ran to 1.973 in seven minutes on liquidations, then went to 2.031, overshooting the calculated target. I took half at 1.968 and trailed the rest behind levels, exiting at 2.018.

The takeaway from this case. Geometry gave the direction and the target. The order book explained why the move stalled at first. The liquidation map explained where the fuel for the burst came from. On the chart alone this trade would have looked like an ordinary breakout, and holding it to 2.018 would have been terrifying.

When the Pattern Doesn't Work

There are three market states in which head and shoulders gives no edge at all. It's easier to memorize them than to find out each time with your own account.

A strong higher timeframe trend. A classic pattern on 15 minutes inside a powerful 4-hour rally plays out a third of the time at best. Formally everything is right: three extremes, a neckline, the volume profile. In reality it's a pullback inside the trend, and after the break price collects stops for 0.3-0.5%, turns around, and goes to make a new high.

A heavy block of resting limit orders behind the neckline. We talked about this scenario above, but let me repeat it with numbers: if the density level's volume is two to three times the average cluster volume on that coin, the break will almost certainly choke. The pattern is drawn, the stops are sitting under the line, but there's nothing to reach them with.

The period before major news. An hour before inflation data or a rate decision, the market compresses, volume drops, and any geometry stops working. The release throws price in an arbitrary direction, and nobody cares about your markup.

Let me break down a failure with specifics. ETH/USDT, hourly chart, uptrend on the 4-hour. A nice classic pattern came together: head at 3,480, neckline at 3,395, right shoulder at 3,452. Calculated target 3,310.

The break happened, the bar closed at 3,382. Seemingly all by the book. But the skew on the tape was sluggish, there were no large sells, just small market orders of 2-5 ETH. A buy-side density level formed almost immediately in the book below the neckline, around 3,370.

Then came the standard false-break scenario. Price poked into 3,368, collected the stops of everyone who went long from the neckline, and within two hours was back at 3,430. The short at 3,380 with a stop above the right shoulder closed in the red.

What was visible in advance. First, the higher timeframe direction was against the trade. Second, the tape didn't accelerate on the break. Neither of those signals requires complicated analysis, both read in ten seconds. That time I ignored them, because the pattern was too pretty.

Mistakes

I've collected the ones people blow up on most often. Some of them I collected on my own account in my first year.

Drawing the pattern in. The most common one. A trader wants to see a head and shoulders pattern and finds it in three random candles. The check is simple. Clear all the drawings off the chart (in the terminal that's the "eye" icon, one click) and look at raw price. If the pattern doesn't read without the lines, it isn't there.

Entering before the break with no reason. An aggressive entry from the right shoulder has a right to exist, but only when there's a separate argument for it: a density level in the book above the shoulder, seller dominance on the tape, absorption on the cluster chart. "The shoulder is about here" is not an argument.

Ignoring the higher timeframe. A bearish pattern on 15 minutes inside a strong 4-hour up move is usually just a pullback that will run you over on the continuation. It doesn't cancel the trade, but it does require cutting size.

A textbook stop with no order book. Above the right shoulder's high can sometimes be 2% of distance. At 0.5% risk of the account that's a position one quarter of your usual size, and most people simply ignore that and keep the usual lot. Risk in that trade goes up fourfold.

Dragging the stop. Price goes against you and the trader moves the stop — "just a little more, the pattern is right after all." The consequence is direct: the loss grows, all the way to liquidation. The stop moves in the direction of profit only. Only.

Treating the target as dogma. The measured move is a reference, not a promise. A large density level on the way outweighs any formula.

Trading the pattern on the 1-minute chart. Too few stops collected, too much noise, fees and the spread eat the whole potential. On a thin instrument, slippage gets added on top.

Forgetting funding rate while holding. The funding rate is a periodic payment between longs and shorts on perpetual futures, usually every eight hours. If a trade on an hourly pattern lives a long time and spans a settlement, the rate can eat a noticeable chunk of the profit. When the rate is negative for your side it's a bonus instead, but you need to count it in advance.

Entering on a wick poke. A wick below the neckline with no close is a stop hunt, not a break. Very often after a poke like that price comes back and goes up, and you're already short.

If clusters and delta are still hard to read, watch the free lesson on market analysis through the order book and cluster charts. We recorded it specifically for beginners; it's part of the same free trading-from-scratch course.

FAQ

• How reliable is the head and shoulders pattern?

  With the filters applied, it plays out around two thirds of the time; without filters it drops to about half. The main filters are alignment with the higher timeframe, volume confirmation on the right shoulder, and the presence of liquidity beyond the neckline. Bare geometry without those conditions gives you coin-flip results.

• Where exactly should I put the stop?

  Beyond the nearest density level in the order book that sits farther than the calculated ATR × 1.5 distance from your entry. The textbook version, "above the right shoulder's high," works on hourly and 4-hour patterns, but for intraday it often gives too wide a stop and kills the risk-to-reward.

• What do I do if the pattern formed but there's no break?

  Nothing. No break, no trade. About a quarter of pretty structures just die into a range, and price then goes up, wrecking the whole markup. Until a bar closes beyond the neckline by at least 0.2 ATR, the signal doesn't exist.

• Is a neckline retest mandatory?

  No, it happens about half the time. Waiting for a retest as a mandatory condition means missing half the working moves. A reasonable compromise is to take part of the position on the break and hold the rest for a possible return to the level.

• Can the pattern be traded on the 1-minute chart?

  Technically yes, practically there's no point. On the 1-minute chart the structure forms out of noise, there aren't enough stops behind it, and the move after the break often doesn't cover the entry and exit fees plus the spread. The minimum working timeframe for crypto is 15 minutes, the optimal one is the hourly.

• How does the inverted pattern differ from the classic one in trading?

  In the volume requirement on the right shoulder and in the character of the move after the break. The inverted one needs rising volume, otherwise the break is false. But when it is confirmed, it overshoots the calculated target more often, because a cascade of short liquidations joins the move.

• How do I tell that a break is false?

  By the tape and the cluster chart in the first minutes after price leaves the level. A real break comes with accelerating trade flow and large prints in the direction of the move. A false one looks like price passing the level on a sluggish tape, after which aggression appears on the other side of the order book and price quickly returns behind the line.

What to Take Away

The pattern by itself gives no edge. The edge comes from understanding that other people's stops are sitting below the neckline, and from checking whether the market has enough aggression to reach them.

The workflow fits into five steps. Find the structure on 15 minutes or higher, on a coin with normal turnover. Check volume element by element and delta on the cluster chart. Draw the neckline through candle bodies with the magnet. Wait for a bar to close beyond the level and for confirmation on the tape. Calculate the stop from the nearest density level, and the position size from a fixed risk amount.

The fifth step matters more than the first four. You can get the markup wrong and still finish the month green if size is calculated properly. The reverse never works.

Watch the pattern and the liquidity in one window.

Secret Terminal puts the chart with its markup tools, the order book with the density map, the tape, and cluster charts on one screen. The magnet snaps the neckline exactly to candle extremes, the ruler (R) measures the head height right inside the order book, auto-stop places protection as soon as your breakout entry fills, and stop and take are moved with the mouse on the chart. No switching between tabs in the one second when the neckline finally gets pierced.