Chart Patterns Tested: Head and Shoulders, Double Bottoms, Triangles, Flags

Chart patterns are shapes that price draws on a chart often enough to have names: head and shoulders, double bottoms, triangles, flags and wedges. Classic textbooks split them into reversal and continuation patterns, and each comes with a confirmation line, a way to measure a target, and a place for the Stop loss. This article gives working definitions, plans a trade step by step, and then looks at what happened when we traded these shapes with fixed rules.

A pattern has three parts, and when it counts

Every pattern in this article has the same frame.

1. The shape — highs and lows lined up in the named form
2. The confirmation line — the line price must close through before the pattern counts, such as the neckline of a head and shoulders or the middle peak of a double bottom
3. The measured move — a target taken from the pattern’s height and projected from the breakout point

Part 2 matters most. Until price closes through the confirmation line, you only have a pattern that might form. The rally that looks like a right shoulder may be heading for a new high. The second low may just be a stop on the way down. That is why most textbooks wait for the candle’s close, not a wick that pokes through.

Head and shoulders, and the inverse

Head and shoulders is a reversal pattern at the end of an uptrend, with three peaks.

• Left shoulder — the first peak in the uptrend
• Head — a higher peak
• Right shoulder — a peak lower than the head, roughly level with the left shoulder (it doesn’t have to match exactly)
• Neckline — the line through the two lows, one between the left shoulder and the head, one between the head and the right shoulder. It can slope; it doesn’t have to be horizontal

The pattern is confirmed when a candle closes below the neckline. The target comes from the head height: the vertical distance from the top of the head down to the neckline (measured below the head), projected down the same distance from the breakout point.

After the break, price sometimes comes back up to touch the neckline from below before falling further. That is a retest, an example of role reversal, but it doesn’t always happen.

Inverse head and shoulders is the mirror image in a downtrend: three lows with the middle one deepest. It is confirmed by a close above the neckline drawn through the two peaks between them, and the target is measured upward the same way.

There are two common places for the Stop loss.

• Above the right shoulder — tighter, and if price climbs back above the right shoulder the pattern has failed anyway
• Above the head — much wider. If price gets back above the head, the uptrend is still going

Either way, leave a small buffer beyond the high, at least more than the spread.

Worked example: planning a head and shoulders trade on EURUSD

Using the prices in the figure above, with a 5,000-dollar account risking 1% per trade = 50 dollars.

Step 1: height and target

• Head 1.1120, neckline 1.0980
• Height = 1.1120 − 1.0980 = 0.0140 = 140 pips
• Target = 1.0980 − 0.0140 = 1.0840

Step 2: entry and Stop loss

• The confirming candle closes at 1.0970; sell there
• Stop loss above the right shoulder at 1.1055 plus 10 pips = 1.1065
• Stop distance = 1.1065 − 1.0970 = 95 pips
• Distance to target = 1.0970 − 1.0840 = 130 pips

Step 3: RR and break-even win rate

• RR = 130 ÷ 95 = 1.37
• Break-even win rate = 1 ÷ (1 + 1.37) = 42.2% (formula from the break-even win rate, before spread)

Step 4: lot size

• On EURUSD one standard lot is worth 10 dollars per pip
• Lots = 50 ÷ (95 × 10) = 0.053, rounded down to 0.05 lots. Actual risk is 0.05 × 95 × 10 = 47.50 dollars

Move the Stop loss above the head instead (1.1120 + 10 pips = 1.1130) and everything changes.

Stop loss	Distance	RR to target	Break-even win rate	Lots for 50 dollars of risk
Above right shoulder, 1.1065	95 pips	1.37	42.2%	0.05
Above head, 1.1130	160 pips	0.81	55.2%	0.03

Same pattern, same target, but the safer Stop loss means you must win more than half your trades just to break even. That is why you work it out before you click. The lot calculator and RR calculator do the arithmetic, and position size from your Stop loss explains the method.

Double tops and double bottoms

A double bottom forms when price makes a low, bounces to a peak, then comes back down to test the area of the first low again. The two lows sit at about the same level.

• Confirmation line — the middle peak between the two lows. The pattern is confirmed when a candle closes above it
• Target — the height from the low to the middle peak, projected up from the confirmation line
• Stop loss — a little below the lower of the two lows. Some traders use the second low, which is closer

A double top is the mirror image: two peaks at about the same height, confirmed by a close below the low between them.

The grey dashed path is the heart of the problem. When price reaches the second low, nobody knows yet whether it will follow the solid line or the dashed one. Buying the second low because you “see a double bottom” is a bet on the solid line. Notice too that in this example the Stop loss distance from entry (95 pips) is longer than the distance to target (75 pips). With RR below 1, you need to win more than half the time to break even.

Triangles: ascending, descending, symmetrical

A triangle forms when the lines through the highs and the lows converge. Price swings in a narrowing range until it breaks out one side.

• Ascending triangle — a flat top at an old resistance level, with rising lows. Most textbooks see it as leaning toward an upside break
• Descending triangle — the mirror: a flat bottom with falling highs, seen as leaning toward a downside break
• Symmetrical triangle — falling highs and rising lows together. The shape itself gives no direction; you wait to see which side breaks

Every kind can break either way. “Leaning toward” doesn’t mean certain, so always wait for a candle to close outside the line. The target uses the height of the base (the widest part, where the pattern starts), projected from the breakout point. The Stop loss goes back inside the pattern, beyond the last high or low before the break.

Flags and pennants

Continuation patterns after a strong move.

• Flagpole — the sharp, fast leg
• Flag — a pause in a small, narrow channel tilted slightly against the pole, for example a down-sloping channel after a rise
• Pennant — the pause is a small symmetrical triangle instead of a channel

The pattern is confirmed by a close through the edge of the flag in the direction of the pole. The usual target is the length of the flagpole, projected from the breakout. The Stop loss goes on the other side of the flag. If the pause is so deep or so long that it eats much of the pole, most traders no longer treat it as a flag.

Wedges

A wedge has two lines that slope the same way and converge, unlike a triangle where one line is flat or the lines slope in opposite directions.

• Rising wedge — highs and lows both rise, but the lows rise faster. Price is going up while the swings narrow. Textbooks see it as leaning toward a downside break
• Falling wedge — the mirror image, seen as leaning toward an upside break

A common target is the start of the wedge, or the height of the wedge at its widest. The Stop loss goes beyond the last high or low inside the wedge.

Pattern	Confirmed when	Target (measured move)	Common Stop loss
Head and shoulders	Close through the neckline	Head-to-neckline height	Above the right shoulder, or above the head
Double top / bottom	Close through the middle low or peak	Pattern height	Beyond the more extreme peak or low
Triangle	Close outside either line	Base height	Last high or low inside the pattern
Flag / pennant	Close through the flag edge in the pole’s direction	Flagpole length	Other side of the flag
Wedge	Close outside the line against the slope	Start of the wedge, or its widest height	Last high or low inside the wedge

How often do these patterns fail

Textbooks tend to show the patterns that worked. On real charts you also see:

• False breaks — price closes through the line, then comes back inside the pattern. The Stop loss is hit and only then does price go the expected way
• Falling short — the break goes the right way but runs out of steam before the measured move
• Patterns that never confirm — a right shoulder that never breaks the neckline is never counted as a “failed” head and shoulders. It simply drops out of memory

The measured move is a rule from the classic textbooks, not a probability that has been shown to be reached. Success rates quoted around the web vary widely because each source defines the patterns differently and uses different markets and periods, so they can’t be compared directly. A well-known academic study, Lo, Mamaysky and Wang (2000), used a computer to find shapes such as head and shoulders and double bottoms in US stocks, and found that some of them carried a little extra information about what came next. That is a statistical measurement, not evidence that trading them is profitable after costs.

One more caution for spot Forex: there is no central traded volume. Stock textbooks use volume to help confirm a pattern, but in Forex you only have each broker’s tick volume, which is a rougher stand-in.

The hindsight problem

Look for head and shoulders on an old chart and you will find plenty of beautiful ones, because you already know where price went. Your brain picks the highs and lows that fit and skips the shapes that didn’t follow the book. This is hindsight bias on charts.

When you trade live, every pullback in an uptrend might be a right shoulder, every bounce in a downtrend might be the middle peak of a double bottom, and nobody knows until price closes through the confirmation line. The defence is to write the definition in numbers first, for example how far the right shoulder may differ from the left as a percentage of the pattern height, and then record every pattern that meets it in advance, including the ones that fail. Related traps are covered in the biases behind trading mistakes.

The clearest pattern on a chart is often clear only because you know how it ended. If price hasn’t closed through the confirmation line, treat it as no pattern yet.

In our backtests

We turned the idea of “buying the bottom” into fixed rules on gold (XAUUSD) M1 charts, using real tick data from 1 January to 1 September 2026 in the MT5 Strategy Tester, with a 10,000-dollar starting balance and 0.10 lots per trade. Highs and lows came from the ZigZag indicator at two sizes (large swings and small swings), counting only legs of at least 1% of price. Every level is a percentage of the leg from the high down to the low.

The results came in three stages.

1. Buying while the bottom was still forming — all 12 settings lost money (Profit factor 0.48–0.98, with drawdowns of 65–100% along the way). The reason is simple: a bottom that is still forming isn’t a bottom. Every time price made a new low the system bought again and was stopped out again. Widening the Stop loss only made the losses arrive more slowly
2. Waiting for a bounce, then buying the retest of the low — far fewer trades, but all 24 settings still lost (Profit factor 0.65–0.94). Repeated retests of the low were often a downtrend carrying on
3. The full double bottom — a bounce, a retest of the low, then price had to rise through a confirmation level before entry

Stage 3 lost money in 14 of 16 settings. The two winners had only 20 and 22 trades in 8 months, too few to tell apart from luck. All 16 together lost 20,751 dollars. Confirmation made the losses smaller than in stage 1, but it didn’t turn them into profit.

We also tried another angle. We labelled the shape that formed before each entry — head and shoulders, double tops and bottoms, triangles and wedges — on every trade of one trend-following system (gold M1, January 2022 to August 2026, ticks generated from 1-minute data, 1,867 entries), then checked which shapes reached a far target more often. Across all trades 17.8% reached it; split by shape the figure ranged from about 12% to 22%. That looks like the shape matters, but:

• Splitting the same trades by something meaningless, the minute of entry (remainder after dividing by 6), gave a range of 16.1–20.8%. Gaps of several points appear by pure chance
• The two swing sizes disagreed. Trades where the shape pointed with the trade did better on large swings (19.3% vs 16.0%) but the reverse on small swings (17.4% vs 20.3%)
• The symmetrical triangle, which our earlier count had ranked worst, came out at 22.3% — the top of the table on small swings

So in this test, the name of the shape didn’t separate good trades from bad ones beyond chance.

Limits: one market, a small timeframe, and our patterns are defined from ZigZag turning points, which differs from reading a chart by eye. Patterns on daily charts or in other markets may behave differently. Backtest results do not predict future results, and the figures above are not returns to expect.

Patterns are good for planning: they tell you when the idea is wrong, where the Stop loss goes and how far to aim. They don’t tell you which way price will go.

Combining patterns with risk management

• Work it out before every entry — Stop distance, distance to target, RR and break-even win rate. If the RR to target is below 1, you must win more than half the time. Ask whether you have evidence that you can
• Lot size comes from the Stop distance — a bigger pattern means a wider Stop loss and a smaller lot. The money at risk stays the same
• Let the pattern plan the trade, not pick the direction — combine it with the larger trend and levels other traders also watch, such as trendlines and channels, instead of trading every shape you see
• Record every pattern that meets your rules — including the ones you skipped and the ones that failed. Split results by pattern type before adding real money. On how test results compare with live ones, see backtest vs live results

Common mistakes

• Entering before confirmation — selling while the right shoulder hasn’t broken the neckline, or buying the second low before a close above the middle peak, is a guess that the pattern will form. In our tests this was the version that lost the most
• Counting a wick as a break — use the candle close on the timeframe you trade, not the price during the candle
• Moving the lines to fit — redrawing the neckline after you’ve seen where price went, or calling shoulders that don’t match at all a head and shoulders, makes everything look like it works
• Treating the measured move as a promise — the pattern-height target is a reference distance. Price has no duty to reach it
• Placing the Stop loss by comfort instead of structure — a Stop tighter than the point where the pattern fails gets swept easily, and one far too wide ruins the RR. Put it where being hit means the pattern was wrong, then size the lot to that distance