Chart patterns explained: head and shoulders, double tops, triangles and flags

Chart patterns are recurring shapes that price traces around support and resistance, such as the head and shoulders, double top, triangle and flag. Traders use them to frame a plan: a level that confirms the pattern, a place where it would be proven wrong, and a rough target. They describe what price has already done; they do not tell you what it will do next, and their track record depends heavily on how strictly you define them and how you test them.

This guide covers the best-known reversal and continuation patterns, shows how traders build entries, stops and measured-move targets around them, and spends as much time on their limits and on testing as on the shapes themselves. It builds on support and resistance and how to read candlestick charts, which deal with levels and individual candles.

What a chart pattern is, and is not

Classic chart patterns were catalogued in the first half of the twentieth century, mostly from stock charts, in books by writers such as Richard Schabacker and, later, Robert Edwards and John Magee. The underlying idea is that crowd behaviour leaves recognisable footprints: a market that fails twice at the same high, or makes lower and lower highs into a flat floor, is showing a shift in the balance between buyers and sellers.

That idea is plausible, but three facts keep it in proportion:

• Patterns are drawn after the fact. It is easy to see a perfect head and shoulders on a past chart. In real time, the right shoulder might turn into a new high, and the pattern simply never existed.
• Definitions are loose. How symmetrical must the shoulders be? How flat the neckline? Two traders looking at the same chart often disagree, which makes it hard to measure how often patterns "work".
• Survivorship in examples. Books and social media show clean patterns that played out. The failures are rarely posted.

So treat each pattern as a way to organise a trade plan, not as evidence that the market will move.

Reversal patterns

Head and shoulders

A head and shoulders top has three peaks: a left shoulder, a higher peak (the head), and a lower right shoulder. The lows between the peaks are joined by a line called the neckline, which may be flat or sloped.

• Formation. In an uptrend, price makes a high (left shoulder), pulls back, makes a higher high (head), pulls back to around the same area, then rallies again but fails below the head (right shoulder).
• Confirmation. The pattern is generally treated as complete only when price closes below the neckline. Until then, it is just three peaks.
• Invalidation. A move back above the right shoulder, or for some traders above the head, suggests the pattern has failed.
• Measured move. The conventional target is the vertical distance from the head to the neckline, projected down from the breakout point.

The inverse head and shoulders is the mirror image at the end of a downtrend: three troughs with a lower middle one, confirmed by a close above the neckline.

Double top and double bottom

A double top is two peaks at roughly the same level with a trough between them. It suggests price tested a resistance level twice and failed.

• Confirmation. A close below the trough between the peaks. Without that break, a double top is just a range.
• Invalidation. A move above the peaks.
• Measured move. The height from the peaks to the trough, projected down from the trough.

A double bottom is the reverse: two troughs at a similar level, confirmed by a close above the peak between them. Triple tops and bottoms follow the same logic with three tests.

Continuation patterns

Triangles

Triangles are periods where price swings within converging trend lines.

• Symmetrical triangle: lower highs and higher lows, converging. Often treated as neutral; the direction of the break is what matters.
• Ascending triangle: a flat top and rising lows. Read as buyers pressing against a fixed resistance.
• Descending triangle: a flat bottom and falling highs. Read as sellers pressing against fixed support.

The measured move is usually the height of the triangle at its widest point, projected from the breakout. Triangles that break late, near the apex, are often considered weaker, because by then the swings are small and almost any move breaks a line.

Flags and pennants

A flag is a short, shallow counter-trend channel after a sharp move (the flagpole). A pennant is the same idea with converging lines, a small triangle. They are read as a pause before the trend resumes.

• Confirmation. A break of the flag's boundary in the direction of the prior move.
• Invalidation. A deeper retracement than a flag should allow, often taken as more than half the flagpole.
• Measured move. The length of the flagpole, projected from the breakout.

Worked example: planning a head and shoulders trade

These numbers are illustrative only. Imagine a daily GBP/USD chart:

• Left shoulder peak: 1.2800.
• Head: 1.2900.
• Right shoulder: 1.2810.
• Neckline: a flat line at 1.2700, joining the two troughs.

The head-to-neckline height is 1.2900 − 1.2700 = 200 pips. The conventional measured target is 200 pips below the neckline, 1.2500.

Suppose the plan is to sell only after a daily close below the neckline, and the confirming close is 1.2690. The stop goes above the right shoulder, the level whose break would suggest the pattern failed: 1.2810 plus a 20-pip buffer for spread and noise, so 1.2830. The stop distance is 1.2830 − 1.2690 = 140 pips.

With a 10,000 USD account risking 1% (100 USD), and GBP/USD worth about 10 USD per pip per standard lot:

• Position size = 100 ÷ (140 × 10) = 0.0714 lots, rounded down to 0.07 lots.
• Loss if stopped: about 140 × 0.07 × 10 = 98 USD plus costs.
• Distance to the 1.2500 target: 190 pips, or about 190 ÷ 140 = 1.36 times the risk.

Three observations. The stop is wide because the pattern is large; the position is correspondingly small. The reward-to-risk ratio is modest, so the pattern needs to reach its target fairly often to pay for its failures; risk-reward ratio explains the arithmetic. And the target is a convention, not a forecast. Price may stall at an intermediate support, or overshoot.

A common variation is to wait for a retest of the broken neckline from below before selling, which can allow a tighter stop. The cost is that some breakdowns never retest, so you miss those trades. Which variation suits you is a testing question, not a matter of opinion.

Worked example: a double bottom with a failure

Illustrative again. On a four-hour USD/JPY chart, price makes a low at 148.00, rallies to 149.20, falls back to 148.05, and rallies again. The pattern's confirmation level is 149.20, the peak between the troughs.

Price closes at 149.30, confirming the double bottom on these rules. You buy at 149.30 with a stop below the second trough: 148.05 minus a 25-pip buffer, 147.80. That is 150 pips of risk. The measured target is the height, 149.20 − 148.00 = 120 pips, added to 149.20: 150.40.

With a 10,000 USD account risking 1%, and a USD/JPY pip value of roughly 6.65 USD per standard lot at that price (1,000 yen divided by about 150.4, which you should recompute at the live rate):

• Position size = 100 ÷ (150 × 6.65) = 0.1003 lots, rounded down to 0.10 lots.

Now suppose the breakout fails: price rises to 149.60, then a data surprise drives it back through 148.00 and the stop is hit at 147.80, with a few pips of slippage. The loss is close to the planned 1%, plus costs. Nothing about the pattern was "wrong"; patterns fail regularly, and the plan was built to survive that. The pip value calculator and position size calculator handle pairs like USD/JPY where the pip value moves with the rate.

Why patterns fail

• Breakouts are frequently false. Price breaks a neckline or triangle edge, triggers stops and entries, and reverses. The breakout trading guide covers false breaks in detail.
• News overrides shapes. A central bank surprise or a major data release can erase any pattern within minutes. Check the economic calendar before acting.
• Forex trades around the clock. Patterns drawn on daily candles depend on your broker's daily close time, so different brokers can show different shapes.
• Small patterns are mostly noise. On low timeframes, almost any sequence of swings resembles something, and spread costs take a large bite out of small measured moves.
• Everyone sees the same levels. Obvious necklines and triangle edges attract stop orders, which can make the area around them choppy.

What the evidence looks like

Academic and central bank researchers have tested chart patterns more formally than most trading books do. One frequently cited example is a 1995 Federal Reserve Bank of New York staff report by Carol Osler and Kevin Chang, which used a computer algorithm based on published technical analysis manuals to identify head and shoulders patterns in daily exchange rates of major currencies against the dollar from 1973 to 1994, and compared the results of a trading rule with simulated random-walk data. Studies like this are worth reading for their method: they show what it takes to define a pattern objectively and to ask whether results differ from chance.

Two cautions apply to any such research. First, results from decades-old data say little about how a pattern behaves in today's markets, with different participants, algorithmic trading and lower costs. Second, a statistically detectable effect can still be too small or inconsistent to survive spreads, slippage and the occasional large loss. Be wary of anyone quoting precise success rates for a pattern without stating how it was defined, on which markets and over what period.

How to test chart patterns honestly

Because patterns are subjective, testing them takes more discipline than testing an indicator.

1. Write an objective definition. For a double top, for example: two swing highs within 0.3 ATR of each other, separated by at least 10 bars, with a trough at least 1 ATR below them. Define swing highs precisely, as described in price action trading.
2. Define the full trade. Confirmation (a close beyond the level), entry, stop, target or exit rule, and position size. Write them down before looking at results.
3. Collect every instance, not just the clean ones. Scroll through history bar by bar, ideally hiding the future, and log every pattern that meets your definition, including the ugly ones.
4. Include costs. Spread, commission, swap for multi-day holds and a slippage allowance.
5. Compare with a baseline. Would a simple breakout of a 20-bar range, with the same stop and target logic, have done as well? If so, the pattern adds little.
6. Check out of sample. Test on a period and pairs you did not use to refine the rules.
7. Forward test and journal. Run the rules on a demo account and record each trade in a trading journal, with a screenshot at the time of entry.

The backtesting guide covers sample size, overfitting and how to read results. If after all that the pattern adds nothing over a simpler rule, that is a useful finding too.

Common mistakes

• Trading the pattern before it is confirmed. Selling the right shoulder before the neckline breaks is a guess that the pattern will form.
• Bending the definition to fit. If you have to squint or redraw the neckline three times, it is probably not the pattern your rules describe.
• Treating the measured move as a promise. It is a rule of thumb. Plan exits that also work if price stops short.
• Placing stops too tight. A stop inside the pattern is likely to be hit by normal noise. Put it where the pattern is invalidated and reduce size instead, as where to put a stop loss explains.
• Ignoring the bigger picture. A bearish pattern inside a strong higher-timeframe uptrend carries different odds from one at the end of a long decline. See multiple timeframe analysis.
• Counting only the winners. Memory and screenshots favour the patterns that worked. A journal corrects for that.
• Oversizing on "perfect" patterns. No pattern justifies extra leverage. Leverage magnifies losses, and most retail CFD accounts lose money.

Frequently asked questions

Is the head and shoulders pattern reliable?

Not reliably enough to trade on faith. It is a widely watched pattern that gives a clear structure for entries and stops, but it fails often, and its apparent reliability depends heavily on how strictly it is defined and how results are measured. Test your own definition on historical data with costs before relying on it.

What is the neckline in a head and shoulders pattern?

The neckline is the line joining the lows between the left shoulder and the head, and between the head and the right shoulder (or the highs, in an inverse pattern). A close beyond the neckline is what most traders treat as confirmation that the pattern is complete.

How is a measured move target calculated?

For a head and shoulders, take the vertical distance from the head to the neckline and project it from the breakout point. For a double top, use the distance from the peaks to the trough. For a flag, use the length of the flagpole. These targets are conventions, and price often stops short or goes further.

What is the difference between a double top and a range?

A double top is only confirmed when price closes below the trough between the two peaks. Until then, two failed tests of resistance are simply a range, and price may go on to break above the highs.

Do chart patterns work in forex?

They can be used in forex like any market, and their underlying idea, that price reacts around support and resistance, is reasonable. Whether specific patterns add value on a specific pair and timeframe, after spreads and slippage, can only be answered by objective testing.

Should I trade chart patterns on low timeframes?

Small patterns on one- or five-minute charts are common and mostly noise, and their small measured moves are heavily eroded by spread costs. Many traders find patterns on four-hour and daily charts easier to define and less sensitive to costs, though the stops are wider and positions must be smaller.