A reversal swing tries to catch a turn against the prior move — buying near the end of a downtrend (or shorting near the end of an uptrend) on the premise that the existing trend is exhausting and a new one is about to begin. This is the hardest, lowest-base-rate corner of swing trading because it is inherently counter-trend: you are fading the dominant direction, so the burden of proof is on you, not the market. The defining discipline is therefore to confirm, never anticipate — wait for a structural break (a neckline, a trendline, a higher-low forming) before entering, rather than guessing the exact top or bottom. The unavoidable tension is that the longer you wait for confirmation the more of the move you give up, but entering early means repeatedly catching falling knives in trends that simply keep going.

The setups

Pattern definitions live in Technical Analysis → Chart Patterns; below is the swing application — the operational trigger / entry / stop / target for each.

• Double bottom (W) / double top (M) — two roughly equal lows (or highs) separated by an interim peak (or trough). Trigger is the confirmation break: for a double bottom, a close above the interim peak ("neckline"); for a double top, a close below the interim trough. Stop below the second low (double bottom) / above the second high (double top); target = pattern height projected from the breakout. Bulkowski stresses the pattern is not confirmed until that neckline break.
• Head-and-shoulders top / inverse head-and-shoulders bottom — three peaks (or troughs) with a higher/lower middle ("head") flanked by two "shoulders." Entry on a decisive close beyond the neckline connecting the intervening lows/highs; stop beyond the right shoulder; target = head-to-neckline height projected from the break.
• Rising-wedge / falling-wedge reversal — converging trendlines tilted with the prior move that signal exhaustion. A rising wedge after an advance breaks down; a falling wedge after a decline breaks up. Entry on the breakout (or a retest of the broken line); stop beyond the last swing extreme inside the wedge; target ≈ the wedge's widest height. Often continuation rather than reversal — context decides.
• RSI / MACD divergence reversal — price makes a new high (or low) but the oscillator does not, hinting momentum is fading. This is a condition, not a trigger: enter only on a separate price confirmation (e.g. a trendline break or reversal candle), never on the divergence alone.
• Climax / capitulation reversal — a vertical, high-volume blow-off (selling climax / buying climax) where the last holders panic out. The trigger is not the climax bar itself but a successful test: a return toward the extreme on lower volume that holds, often with a hammer/long-lower-wick close. Stop below the climax low; never buy into the panic.
• Trendline-break reversal — the simplest: a sustained trend breaks its governing trendline, then puts in the first counter-trend swing point (a higher-low after a downtrend, a lower-high after an uptrend). Entry on that first higher-low/lower-high turning up/down; stop beyond the prior extreme.

How it's used in practice

Disciplined reversal traders treat every setup as guilty until confirmed and stack independent signals. A textbook long reversal: a mature downtrend → a structural pattern forming (double bottom / inverse H&S) at a major support level → bullish RSI or MACD divergence into the second low → a confirmed close above the neckline / broken downtrend line, ideally on volume expansion → entry on that close (or a shallow retest that holds). The stop is structural — below the second low or right shoulder — and the target is the measured pattern height, with reward:risk filtered to ≥2:1. Volume is a recurring tell: necklines and breakouts that come on rising volume, and capitulation lows that come on a 2–5× volume spike followed by a quiet retest, are the higher-quality versions. The single most repeated rule across sources is wait for confirmation — the neckline break, the held retest, the first higher-low — rather than trying to pick the exact turn.

Adoption, debate & evidence

This is the honest core. Reversal patterns are widely taught and widely traded, but the evidence is weaker and the realistic odds lower than for trend-continuation — because you are fighting the prevailing trend.

• Reversal is structurally lower-probability. Multiple educators (Investopedia, CME, practitioner guides) note that "many reversal signals fail" and that a trader should never act on a single signal — the confirmed neckline break is what separates a real reversal from a pause. The same data shows the irony that failed reversals (a busted H&S that breaks back the other way) often precede strong continuation, underlining how often the trend wins.
• Bulkowski base rates (measured on real charts). For the head-and-shoulders top, Bulkowski's later large-sample data reports roughly a 19% break-even failure rate, ~16% average decline, and a ~68% pullback rate (over 2,800 patterns) — and he documents that performance suffers when a pullback occurs. (His older 2005 dataset cited a much lower ~4% failure / "rank 1," so quote the figure with its vintage — the numbers drifted as samples grew.) For double bottoms, he reports the pattern is essentially unconfirmed until the neckline break and that throwbacks are very common (e.g. ~67% for the Eve & Adam variant), with the best results when the throwback is shallow (price stays at/above the confirmation line). Treat all of these as idealized "perfect-pattern" figures — live results run worse.
• Divergence is suggestive, not deterministic. This is the most over-trusted signal in the group. Across sources, bearish (or bullish) divergence can persist for weeks while price keeps trending — a stock can print new highs against fading RSI for a long time. Divergence is therefore an edge/condition, not a trigger; trading it naked is a classic way to short strength too early. Require a separate price confirmation.
• Climax/capitulation needs a tested low. Wyckoff-style and IBKR/CMT material is explicit that a selling climax must be successfully tested (a later return on lower, less-intense selling) before it means more than short-term capitulation — "never buy blindly into falling prices."
• Wedges are ambiguous. Practitioner backtests cite a falling wedge breaking up ~68% and a rising wedge breaking down ~63% of the time, but wedges resolve as continuation about as readily as reversal; these figures come from practitioner/vendor studies, not peer-reviewed work — treat as indicative, not established.

Strengths & limitations

• Strengths: when a reversal is confirmed, entry sits very close to the structural invalidation (just beyond the neckline / second low / right shoulder), giving unusually tight stops and large reward:risk; you enter at the start of a potential new trend with the whole move ahead; clear, objective triggers (neckline / trendline break) are easy to define and rule-test.
• Limitations: inherently counter-trend, so the base rate is the lowest of the swing families; the dominant failure mode is the trend simply continuing (the falling knife / busted pattern); divergence persists and tempts early entry; high pullback/throwback rates (H&S ~68%, double bottom ~67%) mean even "good" entries are frequently shaken; Bulkowski's stats are "perfect-pattern" idealizations and skewed by big winners; and everything is regime-dependent — calling a bottom in a strong downtrend, or a top in a strong bull, fights the odds.

Sources

• Thomas Bulkowski, thepatternsite.com / Encyclopedia of Chart Patterns — head-and-shoulders top (~19% break-even fail, ~16% decline, ~68% pullback, 2,800+ patterns; older 2005 ~4%-fail figure noted with vintage); double bottom (neckline-confirmation requirement, ~67% throwback for Eve & Adam, shallow-throwback rule)
• StockCharts ChartSchool — double bottom/top, head-and-shoulders, rising/falling wedge definitions and neckline/breakout mechanics
• Investopedia; CME Group education — reversal-pattern overview, neckline confirmation, "many reversal signals fail / never trade a single signal"
• RSI/MACD divergence sources (TradingSim, Traders Agency, Kavout) — divergence can persist for weeks; suggestive not deterministic; require separate price confirmation
• IBKR Campus / CMT Association (Wyckoff reversal bars); selling-climax guides — capitulation needs a successful test on lower volume before confirmation
• Disputes flagged: reversal base rates are lower than continuation (counter-trend); Bulkowski figures are "perfect-pattern" idealizations and shifted across dataset vintages; divergence is an over-trusted, non-deterministic condition; wedge break-direction percentages are practitioner/vendor estimates, not peer-reviewed