Harmonic Patterns: Gartley, Bat, Butterfly, Crab and the Honest Catch

Harmonic patterns are five-point price structures defined by specific Fibonacci ratios, used to identify a zone where a move may reverse. The ratios are precise, which makes the method feel objective, but choosing which swings to measure is not, and that is where the difficulty lives.

Harmonic Patterns at a glance

What it is and why it works

Harmonic pattern trading grew out of H.M. Gartley’s work in the 1930s and was formalised with specific Fibonacci ratios by later authors, principally Scott Carney. The premise is that price moves in proportional swings, and that certain proportions recur often enough to be worth trading.

Every harmonic pattern uses five points, labelled X, A, B, C and D, forming four legs. XA is the initial move. AB retraces part of it. BC retraces part of AB. CD is the final leg, and where it terminates, point D, is the potential reversal zone, the area where you look to enter against the CD move.

The patterns differ only in the ratios required at each leg. In a Gartley, B retraces 61.8% of XA, and D sits at 78.6% of XA, so D stays inside the original move. In a Bat, B is a shallower retracement of XA, between 38.2% and 50%, and D sits deeper at 88.6% of XA, which makes for a tighter stop just beyond X. In a Butterfly, B retraces 78.6% of XA and D extends beyond X, typically to 127% of XA. In a Crab, the most extended of the group, D reaches around 161.8% of XA, which produces the largest potential reward and the deepest excursion beyond the original move. In all of them, the BC projection into D generally falls between 161.8% and 361.8% of BC, with the more extended patterns requiring the larger projections.

The reason to take any of this seriously is that Fibonacci retracement levels do behave as reference points in real markets, partly because a great many participants watch them, which makes them self-reinforcing. A harmonic pattern is essentially a demand that several such levels agree simultaneously, and multiple independent measurements converging on one price area is a genuinely meaningful thing.

The reason to be careful is equally clear. Nothing tells you which swing highs and lows are the correct X, A, B and C. On any liquid chart there are many candidate points, and a trader looking for a pattern will usually find one. That degree of freedom is what makes harmonic analysis look far more precise than it is.

How to trade it, step by step

1. Work on the 4-hour or daily chart and mark only significant swing points. A valid X, A, B or C should be a turning point that is obvious without zooming: a high or low that produced a real reaction. Using minor wiggles as pattern points is the primary source of invalid patterns, because it gives you enough candidates to fit almost anything.
2. Identify the XA leg first, then measure AB as a retracement of it. Draw a Fibonacci retracement from X to A and see where B landed. Around 61.8% points to a Gartley; between 38.2% and 50% points to a Bat; around 78.6% points to a Butterfly. If B does not land near any of these within a small tolerance, there is no pattern: stop here rather than adjusting your swing points.
3. Check the BC leg. C must retrace between 38.2% and 88.6% of AB, and must not exceed A. This leg is permissive across all the patterns, so it is a filter rather than an identifier, but a C that breaks A invalidates the structure entirely.
4. Project the D point using both required measurements and require them to agree. Measure D as a retracement or extension of XA (78.6% for a Gartley, 88.6% for a Bat, 127% for a Butterfly, 161.8% for a Crab) and separately as a projection of BC. Where those two projections overlap is the potential reversal zone. If they do not overlap closely, the pattern is weak and should be skipped regardless of how it looks.
5. Demand independent confluence at the D zone. Check whether the zone coincides with something you would have marked anyway: a prior swing high or low, a well-tested horizontal level, a supply or demand zone, or the higher-timeframe trend direction. A harmonic pattern supported by independent structure is a considerably better trade than one that exists only in the ratios.
6. Wait for a reaction at D before entering, do not enter on arrival. Price reaching the zone is not evidence of anything; price reaching the zone and stalling is. Drop to a lower timeframe and wait for a clear rejection candle, a failure to extend, or a break of the short-term trend line of the CD leg. Entering blind at D is how traders get run over by a CD leg that simply continues.
7. Place the stop beyond the pattern’s structural invalidation, not at an arbitrary distance. For a Gartley or Bat, that is beyond X. For a Butterfly or Crab, it is beyond the next major extension level past D. Then size from that distance with the position size calculator.
8. Set targets at the standard retracements of the CD leg. The conventional first target is 38.2% of CD, with a second at 61.8%, and the A point as an extended objective. Taking partial profit at the first target and moving the stop is the usual approach, because harmonic reversals frequently produce a reaction to the first target and then fail.
9. Screenshot the pattern before it resolves and log the outcome. This is the discipline that keeps harmonic trading honest. Patterns identified after the move always look excellent; only a record made in advance tells you whether your real-time identification has value.

Size every one of those entries with the position size calculator and check the trade is worth taking with the risk/reward calculator before you commit.

The conditions it needs

Strict ratio tolerances

The ratios are the only objective content in the method, so loosening them removes the constraint entirely. A tolerance of a few percent around each level is reasonable; treating anything between 55% and 70% as "about 61.8%" means you are no longer trading a defined pattern, you are trading a rough zigzag shape.

Confluence with independent structure

The single biggest improvement available. When the D zone coincides with a prior swing extreme, a major horizontal level or a higher-timeframe zone, the trade has a reason to work that does not depend on the pattern being correctly identified. When the pattern is the only evidence, everything rests on your swing selection.

Higher timeframes

Daily and 4-hour patterns use swing points that are unambiguous and take weeks to form, which limits how many candidates exist. Intraday charts generate patterns constantly, most of which are coincidental arrangements of noise.

Trading with the larger trend where possible

Harmonic entries are counter-trend by construction; you are fading the CD leg. A pattern that completes in the direction of the higher-timeframe trend, so that the reversal at D resumes the dominant direction, is materially more likely to work than one that asks the whole market to turn.

Confirmation at the D point

Waiting for price to actually react in the zone converts a prediction into a response. It costs some of the entry price and removes the worst outcome, which is holding a position while an extended CD leg continues indefinitely.

When it fails

• Choosing the swing points is subjective, and that undermines the precision. The ratios look mathematical, but nothing dictates which highs and lows are X, A, B and C. On any active chart there are several plausible sets, and a trader hunting for a pattern will find one. Two competent analysts can label the same chart differently and both produce valid patterns pointing in opposite directions.
• Pattern fits are frequently revised after the fact. When price runs past D, the common response is not to conclude the pattern failed but to re-measure; the Gartley becomes a Butterfly, the Butterfly becomes a Crab, or a different swing point is nominated as X. Because the pattern family covers a wide range of extensions, there is nearly always a re-fit available. That retrospective flexibility is precisely why harmonic trading is hard to trade and easy to teach: the charts in the lesson always worked.
• Point D is a zone, not a price, and price can overshoot it substantially. Entering the moment price touches the level, with a tight stop just beyond, is the most common way harmonic traders lose. Reversal zones are approximate, and the CD leg is often a strong impulsive move that does not respect a single number.
• The pattern completes and price simply continues. A completed harmonic pattern is not a reversal signal, it is a location where a reversal is more plausible than average. A meaningful proportion of valid, textbook patterns fail outright, and any approach that does not budget for that will oversize.
• Counter-trend by construction. Every harmonic entry fades an existing move. Doing that against a strong higher-timeframe trend is a difficult trade regardless of how clean the ratios are, and the pattern itself provides no warning about the wider context.
• Automated scanners produce far too many candidates. Pattern-recognition indicators will populate any chart with harmonic shapes, most of them measured from insignificant swings. Volume of signals is not evidence of edge, and scanner output should be filtered heavily by timeframe, swing significance and confluence before being treated as tradeable.

Which markets this works best on

• EUR/USD: Clean, well-defined swing structure on the daily chart makes swing-point selection less ambiguous.
• Gold (XAU/USD): Produces large, proportional swings that frequently respect Fibonacci retracement levels.
• GBP/USD: Wide, orderly swings on the 4-hour chart give patterns room to develop with clear turning points.
• SPX500 (S&P 500): Deep participation and orderly corrections within trends suit the counter-trend nature of harmonic entries.
• AUD/USD: Well-defined multi-week swings driven by commodity cycles, with legible retracement structure.

For different levels of experience

If you are brand new

Strip away the names and a harmonic pattern is a zigzag with rules. Price moves from X to A, pulls back to B, moves to C, then makes a final leg to D. The pattern only counts if each leg is a specific proportion of another, measured with a Fibonacci tool. If the proportions are right, D is treated as an area where price might turn.

The main patterns differ only in those proportions. A Gartley finishes inside the original move at 78.6%; a Bat finishes deeper at 88.6%; a Butterfly and a Crab finish beyond the start of the move, at roughly 127% and 161.8%. That is genuinely all there is to the classification.

Two warnings before you go looking. First, you decide which highs and lows to measure from, and that choice changes everything, which is why the method feels far more precise than it is. Second, if you learn from examples, notice that every example worked; the patterns that failed are not in the lesson. In live trading a substantial proportion fail.

If you want to try it, use the daily chart, use only obvious swing points, and never enter simply because price reached D: wait until it visibly stalls there. Learning support and resistance first will make you better at harmonics than learning more patterns will.

If your results are inconsistent

The most common intermediate failure is loose tolerances. If you accept anything roughly near the required ratios, you have converted a defined pattern into a vague shape, and vague shapes are everywhere. Tighten the tolerance and take far fewer trades; the constraint is the whole value of the method.

Second, check whether you enter on arrival at D or on a reaction at D. Traders who enter on arrival with a tight stop just past the level get run over regularly, because the reversal zone is approximate and the CD leg is often impulsive. Waiting for a rejection costs a little price and removes the worst outcome.

Third, be honest about re-fitting. If price blows through the D of your Gartley and you immediately re-measure it as a Butterfly with the same X, you are not analysing, you are avoiding a loss. Keep screenshots of patterns identified before resolution and score them. That log will tell you more about your harmonic trading than any additional pattern will, and it is the only defence against the retrospective revision that makes this method look better than it performs.

If you are experienced

The defensible core of harmonic analysis is confluence, not pattern taxonomy. Several independent Fibonacci measurements converging on one price area is meaningful information; the label attached to that convergence is largely decorative. A professional treatment therefore drops the names, computes the projections, and asks how tightly they cluster and whether that cluster coincides with structure derived by other means.

Understood that way, the patterns encode sensible ideas. A Bat completing at 88.6% of XA is a deep retracement that has not broken the origin, structurally, a test of the low that holds, with a naturally tight invalidation just beyond X. A Crab at 161.8% is an extended excursion into an exhaustion region. Both are recognisable order-flow situations that would be identified without any Fibonacci vocabulary at all.

The unavoidable weakness is degrees of freedom in swing selection combined with a pattern family broad enough that a failed fit can nearly always be re-fitted. That makes real-time performance considerably worse than backtested or illustrated performance, and it makes the method resistant to clean evaluation. The disciplined implementation is: fix swing-significance criteria in advance, require both the XA and BC projections to agree within a narrow band, require independent structural confluence, demand a reaction before entry, and size assuming a meaningful failure rate. Harmonics are then a location tool inside a broader process, which is what they are actually good for.

Risk management for this strategy

Harmonic trades are counter-trend entries into a zone rather than a price, which dictates how they must be risked.

Place the stop beyond the pattern’s structural invalidation (past X for a Gartley or Bat, past the next extension level for a Butterfly or Crab) and never at an arbitrary tight distance chosen to improve the apparent risk-reward. The reversal zone is approximate, so a stop placed immediately beyond D will be hit routinely by normal overshoot. Derive the position size from that wider stop using the position size calculator, which means the position will be small.

Budget for failure. Textbook patterns fail regularly, so risk per pattern must be small enough that several consecutive failures are unremarkable. Never add to a losing harmonic position to defend the pattern, and never re-measure the pattern to move the stop; a broken pattern is a closed trade.

Take partial profit at the first CD retracement target. Harmonic reversals frequently produce a reaction that reaches the initial target and then resumes the prior direction, so converting part of the position and reducing risk on the remainder is a better fit for the method’s actual behaviour than holding for the full extended target. Use the risk-reward calculator to confirm the trade is worth taking before you enter. See risk management.

Where Market Structure Pro fits

The hardest part of trading a harmonic pattern is what happens at point D. Price arrives at your zone and you have to decide, quickly, whether the market is genuinely rejecting the level or simply pausing inside a strong continuation. Get it wrong and you are counter-trend in an impulsive move.

Market Structure Pro is useful at exactly that moment because it answers a different question from the one the pattern answers. The pattern tells you where a reversal would be proportionate; MSP tells you what the structure is actually doing on the closed bar: a single verdict of TRADE, TRANSITION or NO TRADE, with a confidence percentage, an A/B/C grade and a plain-English explanation. If price is at your D point and the structure still reads as an intact trend against you, that disagreement is information worth acting on.

Its ranging and chop filter is relevant too, because a large share of intraday harmonic patterns form inside directionless conditions where the swings being measured are noise rather than structure. A NO TRADE verdict in that state is a reasonable prompt to skip a pattern that a scanner has flagged.

Most importantly for a subjective method, MSP does not repaint and locks its state on the closed bar. The reading at the time you took the pattern is the reading you review afterwards, which means it cannot be quietly re-fitted the way a pattern can. For anyone trying to work out whether their real-time harmonic identification adds value, that fixed second opinion is more useful than another indicator. MSP is decision support: it does not identify harmonic patterns, it does not place trades, and it guarantees nothing.

Frequently asked questions

What are harmonic patterns in trading?

Harmonic patterns are five-point price structures, labelled X, A, B, C and D, in which each leg is a specific Fibonacci proportion of another. The point where the pattern completes, D, is treated as a potential reversal zone. The main patterns (Gartley, Bat, Butterfly and Crab) differ only in the ratios they require.

What are the Gartley pattern rules?

In a Gartley, B retraces approximately 61.8% of the XA leg, C retraces between 38.2% and 88.6% of AB without exceeding A, and D completes at approximately 78.6% of XA. Because D stays inside the original XA move, the structural invalidation is a break beyond X.

What is the difference between a Bat and a Gartley?

The B point and the D point. In a Bat, B is a shallower retracement of XA, between 38.2% and 50%, and D completes deeper at 88.6% of XA. In a Gartley, B is around 61.8% and D completes at 78.6%. The Bat's deeper D sits closer to X, which allows a tighter structural stop.

What is a Butterfly pattern?

A Butterfly has B retracing around 78.6% of XA, with D extending beyond the origin of the move, typically to about 127% of XA. Because D lies outside the XA range, the invalidation level sits further out, at the next major extension beyond D.

What is a Crab pattern?

The Crab is the most extended harmonic pattern, with D completing at approximately 161.8% of XA. It offers the largest potential move from the reversal zone but requires price to travel well beyond the original XA range, which makes the invalidation distance larger.

Are harmonic patterns reliable?

They are a location tool rather than a reliable signal. The ratios are objective, but choosing which swing highs and lows to measure is not, and a substantial proportion of correctly measured patterns fail outright. They perform considerably better when the completion zone coincides with independent structure such as a prior swing extreme or a major level.

How do you trade the D point?

Wait for price to reach the zone and then show a reaction (a clear rejection candle, a failure to extend, or a break of the CD leg's short-term trend line) rather than entering the moment the level is touched. The stop goes beyond the pattern's structural invalidation, and the first target is conventionally a 38.2% retracement of the CD leg.

Why do harmonic patterns fail so often?

Chiefly because point D is an approximate zone rather than a precise price, and because a completed pattern indicates a plausible reversal location rather than a reversal. Harmonic entries are also counter-trend by construction, so a pattern that completes against a strong higher-timeframe trend faces poor odds regardless of how well the ratios fit.

Are harmonic patterns good for beginners?

Not particularly. The ratios are easy to learn but the swing-point selection that drives everything is a matter of judgement, and beginners tend to find patterns wherever they look. Learning to read support, resistance and trend structure first produces better results and makes any later harmonic work more reliable.

Related reading

• Elliott Wave Theory: The other major Fibonacci-based framework, with the same subjectivity issue at its core.
• Support and Resistance: Confluence with independent levels is the biggest single improvement to harmonic trading.
• Supply and Demand Zone Trading: A structural way to confirm whether a completion zone is worth trading at all.
• Candlestick Patterns: The reaction at point D is usually read from candle behaviour on a lower timeframe.
• Risk Management: Patterns fail regularly, so sizing must treat failure as ordinary.