Double top pattern in crypto: how to confirm and measure an M

Identify the two peaks and the intervening low

A double top is a two-peak reversal candidate that can form after an advance. The reaction low between the peaks defines the neckline; similarity alone does not confirm a breakdown.

Draw the neckline through the swing low between the peaks. It can slope; use the line’s value at the moment you test for a break. A broad range with several comparable highs may be better described as resistance or consolidation than a clean two-peak pattern. Naming the shape should help organize the evidence rather than force ambiguous price action into an M.

A repeated high can form part of several different structures. Check the number and order of reaction highs before calling it an M:

Some historical classifications distinguish an Adam peak, which is narrow and pointed, from a wider, rounded Eve peak. Adam-and-Adam (A-A) and Eve-and-Eve (E-E) double tops are separate groups in ThePatternSite. They are not interchangeable with every pair of crypto highs, and their published rates should remain attached to the subtype studied.

Define confirmation and invalidation

A downside confirmation can be defined as a trade below the neckline, a completed close below it, or a close followed by a failed retest from underneath. A wick can trigger the rule earlier, but price may return above the neckline before the candle closes. A close rule is easier to reproduce but can put the entry farther from the level. A retest rule may miss a clean move that never revisits the line.

Before confirmation, an advance through the second peak weakens the bearish interpretation. After a neckline break, a close back above the line can mark a failed breakdown under a close-and-hold method. The high of the second peak is a broader structural reference, but a trade’s protective exit may be placed elsewhere to control risk. Keep the pattern’s invalidation rule separate from the account-level stop and position size.

Volume may provide context around each test and the breakdown, but it is not a substitute for price confirmation. Crypto volume can represent spot, perpetual futures, a single exchange or an aggregate; state which feed is shown. A volume spike does not by itself prove selling dominance or a lower chance that price will recover above the neckline.

Calculate a synthetic measured-move reference

Suppose a synthetic chart has two peaks near 120 and an intervening neckline at 110. The pattern height is 120 − 110 = 10. After a later close below the neckline at 108, projecting that height down from the neckline gives a reference near 100; anchoring the full distance from the 108 close gives 98. State the convention you use rather than moving the target after the result is known. Neither number is a forecast.

Look for potential support and liquidity between the breakout and the projection. If a prior swing low sits at 104, the geometric target near 100 may be less relevant than how price behaves around 104. Also compare entry-to-stop distance with the realistic path to the target. A pattern with a large measured move can still lose money on average if losses are frequent or costs are high.

Any performance review needs the full candidate set: two-peak structures that never break the neckline, false breaks, patterns that reach a target only after first crossing a stop and patterns that do neither within the chosen horizon. Record the market, venue, interval, exact pattern rule, entry, stop, target, fees, slippage and (for perpetuals) funding.

Compare historical results by shape and endpoint

Study scope: the figures in this section come from ThePatternSite’s bull-market results for historical stocks. Bull market describes the source’s market regime; it does not mean that every individual pattern or breakout was bullish. These figures do not combine all market regimes.

ThePatternSite’s current historical stock pages report the following separate samples. The site defines break-even failure as a downside breakout that does not fall at least 5% in the expected direction; pullback is another price event, and meeting the measure-rule target is a separate endpoint.

The table deliberately keeps failure and target percentages separate. The A-A group has a higher target-attainment figure in this summary despite a higher 5% failure figure than E-E. A single “success rate” would hide that difference.

These are historical stock-chart results, not a crypto sample, not a validated 15-minute or 4-hour strategy, and not net returns after execution. A 5% favorable move can occur before a reversal and does not ensure a trader exited profitably. The figures also do not tell you how often a trade reaches a target before a stop under your own entry rule. No crypto-specific double-top expectancy is asserted here.

altFINS subtracts a 25% historical stock break-even-failure rate from 100% and calls the remaining 75% a success rate. That 75% means price fell at least 5% after the breakout; it does not measure profitable trades. The same page quotes a 55-65% crypto range after volume confirmation but does not identify its study in the visible text. These are different endpoints and populations.

Common failure cases

Price breaks above the second peak. The expected ceiling did not hold. Do not retain a short only because the chart still resembles an M after that level is exceeded.

Price briefly falls below the neckline, then recovers. A quick trade below the line may not satisfy a close-based rule. If you use wick breaks, capture those that reverse just as consistently as those that continue.

The M is inside a larger uptrend. A lower-timeframe double top can mark a pause or retracement rather than a broad reversal. Higher-timeframe context can change the interpretation but does not guarantee an outcome; the bars overlap and are not independent confirmations.

The projected target is below nearby support. A measured move ignores real-time obstacles. Map prior lows and account for volatility and market liquidity before using its level as an exit reference.

Frequently asked questions