Chart Patterns Reference: Triangles, Flags, Head-and-Shoulders and What Works

Chart patterns are not magic. They are the visual fingerprint of predictable crowd behaviour – accumulation before a breakout, profit-taking that forms a flag, exhaustion that prints a head-and-shoulders. Markets are made up of humans (and algorithms replicating human playbooks), so when similar conditions recur, similar price structures recur with them.

The modern scientific study of patterns began with Richard Schabacker (Technical Analysis and Stock Market Profits, 1932) and was systematised by Robert Edwards and John Magee in Technical Analysis of Stock Trends (1948). Thomas Bulkowski's Encyclopedia of Chart Patterns (1st edition 2000, 3rd edition 2021) ran statistical analyses over thousands of historical patterns on US stocks to quantify breakeven failure rates, measured-move completion rates, and performance in bull versus bear markets. Bulkowski's work remains the closest thing to an empirical benchmark the field has.

Academic consensus is modest. Lo, Mamaysky & Wang (Journal of Finance, 2000) used kernel regression to algorithmically detect patterns in US equity data and found statistically significant, though small, information content for several common patterns. Subsequent studies (e.g., Dawson & Steeley 2003 on UK data) confirm the effect weakens after transaction costs. The patterns work; they are just not easy money.

Chart patterns fall into three families:

Reversal patterns – signal the end of the prior trend. The major reversals are head-and-shoulders (top), inverse head-and-shoulders (bottom), double top, double bottom, triple top, triple bottom, and rounding bottom.

Continuation patterns – pause within an existing trend that typically resolves in the trend direction. Main continuations are flags, pennants, ascending/descending triangles (with directional bias), and rectangles.

Bilateral patterns – can resolve either way, so they are setups for a breakout, not a directional call. Symmetrical triangles, wedges (when in consolidation), and broadening formations are the main examples.

A useful mental model: a continuation pattern is the market catching its breath; a reversal pattern is the market changing direction; a bilateral pattern is the market compressing energy before releasing it somewhere. The trader's job is to identify which family you are in before the breakout tells you.

The head-and-shoulders is a three-peak pattern where the middle peak (head) is higher than the two adjacent peaks (shoulders). A trend line connecting the lows between the peaks forms the neckline. Entry signal: close below the neckline. Measured target: distance from neckline to head, projected downward from the breakout.

Bulkowski's stats (3rd edition, US equities 1990–2019): the classical head-and-shoulders top has a breakeven failure rate of ~7% (i.e., 93% of patterns reach some further decline after breakout), with a meeting-target rate (pattern reaches full measured move) around 55%. Those numbers fall in bear markets (measured-move completion drops to ~40%). In FX majors and index futures, the pattern tends to be messier – liquidity means fewer clean symmetrical formations.

Three common mistakes: (1) entering before the neckline breaks – shoulders fail all the time and become triple tops or consolidation; (2) ignoring volume – Magee called for volume on the first shoulder, lower volume on the head, and higher volume on the neckline break, a criterion retail often skips; (3) treating the inverse head-and-shoulders (bottom) identically – Bulkowski's data suggests bottom reversals complete more often than tops.

Ascending triangle: horizontal resistance with rising support. Buyers repeatedly defend higher lows against a fixed ceiling of sellers. When the ceiling breaks, the pent-up buying typically produces a clean breakout. Bulkowski ranks it among the top continuation patterns in bull markets.

Descending triangle: horizontal support with falling resistance. Mirror image – usually a bearish continuation in a downtrend, though in some bull markets it resolves to the upside (illustrating the need to combine pattern with broader trend context).

Symmetrical triangle: converging trend lines with no clear directional bias. Price makes lower highs and higher lows. These are close to 50/50 in direction – Bulkowski reports roughly 54% of symmetrical triangles in bull markets break up, so the bias is slight. Trade with a breakout trigger rather than pre-positioning.

For all three, volume should decline during the formation (coiling) and surge on the breakout. A breakout on weak volume is the single best signal of a false break. Position only after the breakout bar closes beyond the line, or after a retest holds.

Flags are short-duration parallel-channel consolidations after a sharp move. A bull flag is a downward-sloping rectangle after a fast rally; a bear flag is an upward-sloping rectangle after a fast decline. Typical duration: 5–20 bars on a daily chart, much less on intraday charts.

Pennants are small symmetrical triangles after a sharp move – essentially a flag with converging rather than parallel lines. Duration is similar.

Both patterns are strongly continuation-biased. Bulkowski reports the bull-flag continuation rate around 67% with meeting-target around 60% in bull markets. The pattern works because it represents profit-taking that is quickly overcome by continued buying pressure – the sharp move indicates institutional flow, not a fluke.

Measured move for flags is typically the flagpole – the initial impulse from the start of the move to the start of the flag – projected from the breakout. A tight stop below the flag low (for bull flags) gives a favourable risk-reward, often 1:3 or better.

The chart below plots approximate meeting-target rates (how often the full measured move is achieved after a breakout) for ten common patterns in US equities bull markets, drawn from Bulkowski's 3rd edition ranges. Rates below 50% do not mean the pattern is unprofitable – they mean the pattern typically delivers only part of the measured move. They all sit roughly within 50–70%, which is a more realistic range than the 80%+ numbers quoted in retail pattern tutorials.

Two takeaways: (1) even the "best" patterns miss their full target about 40% of the time, so partial profit-taking is essential; (2) the difference between the best and worst patterns is only about 15 percentage points, far smaller than trading tutorials suggest. Risk management matters more than pattern selection – and that discipline is what carries a pattern into a simulated evaluation.

The cup and handle was popularised by William O'Neil in How to Make Money in Stocks (1988) as part of his CAN SLIM methodology for growth-stock investing. Structure: a rounded, U-shaped base (the cup) lasting weeks to months, followed by a smaller downward drift (the handle) of 1–4 weeks, then a breakout above the handle's high on volume.

The pattern is a trend-continuation structure for stocks in long-term uptrends. O'Neil's requirement was explicit: cup depth no more than 30% from left-side peak to bottom (less in weak markets), handle drift no more than 12–15% down from the right-side rim, volume at least 40% above average on the breakout bar.

Bulkowski's 3rd-edition data places cup-and-handle meeting-move rate near 58% in bull markets, making it one of the more reliable continuation patterns on equities. It translates poorly to FX majors (which have neither growth earnings nor the weeks-to-months base-building behaviour), so treat it as a stock/ETF pattern primarily.

Rising wedge: converging lines both sloping upward, with the upper line rising more slowly than the lower. Typically a bearish reversal when appearing at the end of an uptrend, or a bearish continuation when appearing as a counter-trend bounce inside a downtrend. The pattern works because the buying that is producing higher highs is losing strength relative to the buying that is producing higher lows.

Falling wedge: converging lines both sloping downward, with the lower line falling more slowly. Mirror image – bullish reversal at downtrend bottoms, bullish continuation during uptrend pullbacks.

Wedges are context-sensitive. The same structural shape means opposite things depending on where it appears in the larger trend. A rising wedge inside a bull trend is less reliably bearish than a rising wedge at the top of an extended rally. Always read wedges in conjunction with the trend structure one timeframe higher.

Break targets for wedges are typically measured from the widest part of the wedge, projected from the breakout. Stops go beyond the opposite side of the wedge, which often creates wide stops – wedges are low-leverage setups unless you can enter on a retest.

Double top: price reaches a high, pulls back, rallies to approximately the same high, then fails. Confirmation: break of the intermediate low between the two peaks (the "neckline"). The pattern's logic is simple – sellers successfully defended the high twice, buyers have used their energy without breaking through, control is shifting.

Double bottom: mirror image, used for trend reversals after a decline. Often more reliable than double tops because uptrends in equity markets have secular tailwinds (long-term economic growth, earnings compounding) that bottoms tend to respect more than tops.

Triple top / triple bottom: a third test of the level. Statistically similar performance to double versions; the third failure adds confirmation but lengthens the pattern, widening stops. Bulkowski's data gives triple tops slightly higher meeting-move rates than double tops, at the cost of fewer setups.

Key rule: the two (or three) peaks should be separated by at least a few bars of visible pullback, not adjacent wicks. A "double top" made of two highs four bars apart is just noise. Real double tops usually span 20–60 bars on daily charts.

Rule 1 – Pre-define the pattern before the breakout. Draw the structure and neckline before the break, not after. Post-hoc pattern identification is the #1 source of false confidence.

Rule 2 – Require a clean breakout. A full bar close beyond the breakout level, not a wick. On daily charts, that means wait for the daily close.

Rule 3 – Check volume on the breakout. A breakout on weak volume is the best early warning of a false break.

Rule 4 – Size by stop distance, not by conviction. A triangle with a 50-pip breakout and a 40-pip stop needs smaller position size than a flag with 15-pip stop. Use a position sizing calculator.

Rule 5 – Pre-commit to a partial exit. Bulkowski's data shows 40%+ of patterns miss their full measured move. Taking half profit at the halfway mark captures a large fraction of winners with less regret.

Rule 6 – Respect the retest. A breakout that immediately retraces to the neckline and holds is often the best entry. A breakout that re-enters the pattern without a retest is a flag for a false break.

Rule 7 – Ignore pattern names in ranging markets. Every pattern "works" differently when the market is choppy. If the higher-timeframe is sideways, pattern trading produces more false signals.