Harmonic Patterns: Gartley, Bat, Butterfly & Crab Explained
Harmonic patterns are XABCD price structures where each pivot must hit a specific Fibonacci ratio. A valid Gartley places B at exactly 61.8% of the XA leg. A valid Bat places D at 88.6%. The patterns identify a potential reversal zone (PRZ) where price has historically reversed. You trade the reaction at the PRZ, not the pattern itself.
I spent three years dismissing harmonic patterns as chart-reading numerology. Every “completed” pattern posted online had been drawn after the reversal. Hindsight bias packaged as analysis.
What changed my view was applying strict ratio rules: not “approximately 61.8%” — exactly 61.8%, ± 5%, with a body close confirmation. When I applied that filter to EUR/USD 4H over eight months, the signal quality improved significantly. The patterns I had dismissed were the loosely-drawn ones. The setups following precise ratios had real, measurable edge.
What Makes a Pattern Harmonic
The term “harmonic” refers to proportional relationships. In trading, these are specific Fibonacci ratios between each leg of the price structure.
Every harmonic pattern starts with a baseline move called the XA leg. All subsequent pivots (B, C, and D) must retrace or extend XA by specific percentages. Miss any ratio by more than 5%, and the pattern is invalid. That strictness is the filter that separates harmonic analysis from free-form “it kind of looks like an M” chart reading.
To find these patterns you need a Fibonacci retracement tool and a Fibonacci extension tool. On TradingView both are in the drawing panel. You measure B against XA, C against AB, and D against XA. Numbers align: potential entry. Numbers miss: move on.
The Fibonacci levels used in harmonic patterns (61.8%, 78.6%, 88.6%, 127.2%, 161.8%) are the same ones covered in the Fibonacci trading guide. If those ratios are unfamiliar, read that first.
The ABCD Pattern: Foundation of All Harmonics
The ABCD is the simplest harmonic structure. Four points, no X origin.
In a bullish ABCD, price drops from A to B, then retraces up to C (61.8% to 78.6% of the AB leg). Price then drops again to D, completing at either 127.2% or 161.8% extension of BC. That D completion zone is the PRZ.
The bearish ABCD is the inverse: AB is a rally, C is a retracement, D is the extension beyond B.
Every five-point harmonic pattern builds on this structure. Understanding the ABCD first makes Gartley, Bat, and the others easier to read. Before trading any of the more complex patterns, I worked ABCD setups exclusively for three months. The measurement mechanics transfer directly to all the XABCD patterns.
The Five Main Harmonic Patterns
Each pattern has specific Fibonacci tolerances at every pivot point. These are not approximations.
| Pattern | B point | D point | D relative to X |
|---|---|---|---|
| Gartley | 61.8% of XA | 78.6% of XA | Stays between X and A |
| Bat | 38.2%–50% of XA | 88.6% of XA | Stays between X and A |
| Butterfly | 78.6% of XA | 127.2%–161.8% of XA | Extends beyond X |
| Crab | 38.2%–61.8% of XA | 161.8% of XA | Extends far beyond X |
| Cypher | 38.2%–61.8% of XA | 78.6% of XC retracement | C extends beyond A (113%–141.4% XA) |
Gartley Pattern
Harold Gartley described this structure in 1935. B sits at 61.8% retracement of XA. D completes at 78.6% retracement of XA, with the CD leg extending 127.2% to 161.8% of BC.
I tracked Gartley completions on EUR/USD 4H over eight months of live analysis. Of 23 setups where D closed within the PRZ with candle body confirmation, 14 reversed cleanly — a 61% reversal rate. The failures shared one consistent factor: B missed the 61.8% level by more than 7%. Tight ratio adherence is not optional. It is the edge.
Stop loss: just below D for bullish, just above D for bearish. Target 1: the C level. Target 2: the A level.
Bat Pattern
Scott Carney developed this in 2001. B retraces only 38.2% to 50% of XA (shallower than the Gartley), and D sits at 88.6% retracement of XA.
The 88.6% D level produces precise stop placement. On a bullish Bat, the stop goes just below D. Because 88.6% is a specific cluster rather than a round number, price tends to either bounce cleanly or accelerate through. There is rarely a prolonged hang at that level.
On GBP/USD 4H, the Bat has produced the strongest harmonic results in my live tracking. The 88.6% D level allows a tight stop relative to the expected reversal distance toward C and A, producing average R:R of 2.3 to 2.4 on completed setups. That is a measurable improvement over the Gartley’s 2.1 average on the same dataset.
Butterfly Pattern
The Butterfly places D beyond X, at 127.2% or 161.8% extension of the XA leg. B sits at 78.6% of XA (deeper than the Gartley’s 61.8%). C retraces 38.2% to 88.6% of AB. D then extends XA significantly.
This is counterintuitive. You are entering after price has moved past the prior swing, not before it reaches it.
The logic: if price breaks a major swing and halts at exactly 127.2% extension with Fibonacci ratios confirming the BC leg, that extension is likely exhaustion rather than continuation. That is the trade hypothesis.
I use Butterfly setups only when the weekly chart structure supports a reversal. Without that higher-timeframe context, extension patterns generate too many false positives in trending conditions.
Crab Pattern
The Crab extends to 161.8% of XA at D, the deepest completion in standard harmonic patterns. CD must measure 224% to 361.8% of BC.
Carney considers the Crab the most precise harmonic structure. When the ratios align tightly, the PRZ is narrow despite the extreme extension. That concentration matters: a tight PRZ means a closer stop and better R:R if correct.
I trade Crab setups rarely and only on XAU/USD daily, where the move size justifies the wider initial risk. One textbook bullish Crab on XAU/USD daily caught the pullback entry before a significant upside extension in 2025. Single data point, but the structure was textbook.
Cypher Pattern
The Cypher places C beyond A, specifically at 113% to 141.4% extension of XA. No other standard harmonic places C above the A level. D then retraces 78.6% of the full XC leg.
Cypher setups appear in higher-momentum environments where a strong impulse overshoots. The D retracement of XC is the reversal point. Rarer than Gartley and Bat, but with comparable reversal probability when ratios are strict.
- Each pattern gives a defined entry zone (PRZ), not a discretionary drawing
- Stop placement is unambiguous: just beyond the D completion level
- Targets are pre-existing Fibonacci levels (C and A) requiring no interpretation
- Pattern invalidity is detectable early: a ratio miss over 5% signals abort before entry
- Applicable to forex, gold, index CFDs, and crypto on liquid timeframes
How to Trade Harmonic Patterns: Step by Step
Most traders abandon harmonic analysis because they never apply strict ratio rules. The identification process rewards patience, not speed.
Step 1: Find the XA leg. Look for a clean impulsive move with minimal retracement during the leg. No pullback deeper than 38.2% while it is forming. Mark X and A. This is your baseline for all subsequent measurements.
Step 2: Wait for B. Price reverses and begins retracing XA. Check which Fibonacci level it stalls at with a body close: 61.8% suggests Gartley, 38.2%–50% suggests Bat, 78.6% suggests Butterfly. Mark B only after a candle body closes at that level. Wicks do not count.
Step 3: Measure the BC leg. C must retrace 38.2% to 88.6% of AB for most patterns. Mark C on confirmation.
Step 4: Project the PRZ ahead of price. Use a Fibonacci extension from X to A, anchored at B, to project where D will complete. This gives you a price level to watch before price arrives. Having the level pre-marked removes hesitation at the moment of entry.
Step 5: Wait for price to reach D and require confirmation. Price touching the PRZ is not an entry signal. You need: a bearish or bullish engulfing candle body close, a pin bar, or a clear rejection wick at the PRZ. The reaction at D is the trade. The pattern is the setup map.
Step 6: Set stop loss and targets. Stop: 5 to 10 pips beyond D (or one ATR on daily charts). Target 1: C level. Target 2: A level. Move stop to breakeven after Target 1 is hit.
Entry levels, stop losses, and lot sizes. Updated every trading day. Join free.
Performance Data: What the Numbers Show
Running strict harmonic analysis across EUR/USD 4H and GBP/USD 4H for eight months produced these results on setups where D closed within the PRZ:
| Pattern | Setups identified | Reversals confirmed | Win rate | Average R:R |
|---|---|---|---|---|
| Gartley | 23 | 14 | 61% | 2.1 |
| Bat | 11 | 7 | 64% | 2.4 |
| Butterfly | 8 | 4 | 50% | 2.8 |
| Crab | 3 | 2 | 67% | 3.1 |
Sample sizes are small: this is live chart monitoring over eight months, not a multi-year statistical backtest. The Crab result at three setups is not statistically significant.
What the data does support: with strict ratio rules, reversal rates in the 60–65% range on major forex pairs are achievable. At average R:R above 2, that produces positive expectancy. The Butterfly’s 50% win rate at 2.8 average R:R is mathematically sound, but psychologically difficult. Entries at D extensions feel like chasing a move that has already gone too far. Most traders abandon the setup before D. This is a trading psychology problem as much as a technical one.
Common Mistakes to Avoid
-
Forcing ratios. B at 55% when you need 61.8% is not close enough. That is an invalid Gartley. Wait for the actual level or discard the setup.
-
Entering at D without confirmation. Pattern identification is preparation, not a signal. Strong trends can push through PRZ levels. Candle confirmation at D is required before any entry.
-
Ignoring PRZ width. The PRZ is a zone: the overlap between the XA target ratio and the BC extension. Entries within a tight cluster carry higher probability than entries at a single Fibonacci line. Wide PRZs (50+ pips apart on 4H) require stronger candle confirmation.
-
Trading harmonics against the dominant trend. A bullish Gartley in a daily downtrend is a countertrend trade. Lower probability. Mark trend direction on the higher timeframe before entering any reversal.
-
Drawing from wrong anchor points. Fibonacci measurements require consistent anchoring at X and A from the same swing structure. Shifting anchor points to force a pattern is the most common bias in harmonic analysis.
For a broader look at chart structures that complement harmonic entries, the chart patterns guide covers continuation and reversal setups across multiple pattern families.
FAQ
What is the best harmonic pattern for beginners?
How accurate are harmonic patterns in forex trading?
What tools do I need to find harmonic patterns on TradingView?
What is the difference between a Bat and a Gartley pattern?
Can harmonic patterns work for crypto trading?
How do I draw the Potential Reversal Zone (PRZ) correctly?
🌍 Our recommended brokers
Some links on this page may earn us a commission — at no extra cost to you.
Reader Reviews
I applied the six-step trading process from this article to GBP/USD 4H over ten weeks and tracked every qualified setup in a spreadsheet. Step 4, projecting the PRZ before price arrives, changed how I approach these patterns. I had been reactive before, identifying patterns only after D completed and then making an entry decision under pressure. Pre-marking the D zone removed that hesitation. Of 14 setups where the projected PRZ matched a completed D within 5%, 9 reversed cleanly. Monthly account return averaged 6.8% at 1.2% risk per trade on a $2,800 Exness account. The stop placement detail - 5 to 10 pips beyond D rather than at D exactly - fixed a recurring problem I had. I had been placing stops at the D level itself, which caused three early exits on valid setups where price briefly spiked past the body close before reversing.
The PRZ overlap explanation resolved something I had been doing wrong for months. I had been treating a single Fibonacci level at D as the entry zone rather than the overlap of two converging levels. Once I required both the XA retracement and the BC extension to cluster within 15 pips, my Gartley win rate on EUR/USD moved from 44% to 63% over the next eight weeks.
The Bat section with the 88.6% D level data makes the R:R improvement easy to understand. A tighter cluster at 88.6% means a closer stop, so even with the same reversal probability the numbers improve versus the Gartley. Running Bat setups on GBP/USD 4H for four months produced an average R:R of 2.3 on 9 qualifying setups, which matches the 2.4 figure closely. The one area I found less developed is timeframe selection - I had to test 4H and daily setups separately before the quality difference became clear.
Running the 5% ratio tolerance rule from this article changed my Gartley results immediately. I had been allowing B at 55-58% retracement on EUR/USD and treating those as valid. After switching to strict body-close confirmation at 61.8%, my valid Gartley count dropped from 31 to 19 setups over six months. Reversals confirmed at 63% on the filtered setups versus 48% before. Monthly account return averaged 7.4% at 1% risk per trade on a $3,600 Exness account during those six months. The comparison table with exact ratios per pattern is what I reference most now. Before this I had confused Gartley and Bat B levels, treating 50% as Gartley territory when it's actually the upper bound of a valid Bat. That one distinction prevented three entries with wrong stop placement.
The point about requiring candle confirmation at D rather than entering on price touching the PRZ is the most actionable section in this guide. I had two Butterfly setups on GBP/USD where I entered on touch, both ran through the zone, and I took maximum loss. Three subsequent setups where I waited for a body close: all three reversed and hit the C target.
The Butterfly section with the weekly chart filter requirement matches my own testing closely. After three months of trading Butterfly setups on EUR/USD 4H without a higher-timeframe condition, I switched to requiring a potential reversal structure on the weekly chart at the same level before entering. My next 7 qualifying setups produced 5 reversals. Average R:R on those 5 winners came to 2.7, and monthly return averaged 7.2% at 1.2% risk per trade on a $3,100 account over that quarter. The extension entries remain psychologically difficult, but having the weekly chart as a filter removed most of the false positives.
The technical content here is accurate and the ratio data is useful. What I found is that learning the measurement mechanics takes longer than the article implies if you're starting from zero with Fibonacci tools. I spent six weeks practicing before ratio identification felt natural. The data on win rates and R:R is correct in my experience, but the learning curve to get there was longer than a single reading suggests.
The eight-month EUR/USD and GBP/USD tracking data in the performance section matched my own results closely enough to work as a benchmark. I had been running Gartley setups on EUR/USD 4H for six months with a 57% reversal rate before reading this. The difference came from one adjustment: I had been allowing a 10% ratio tolerance rather than 5%. Tightening to strict 5% body-close confirmation dropped my setup count from 27 to 18 over five months but moved reversal rate to 63%. Monthly account return averaged 7.1% at 1% risk per trade on a $2,200 Exness account during that period, compared to 4.3% in the six months before. The PRZ width section is the part most guides skip entirely. Having a threshold for tight versus wide PRZ and requiring stronger confirmation on wide ones reduced my worst losses. I had been treating a 70-pip PRZ the same as a 12-pip PRZ on the 4H chart. That is not the same trade.
Leave a Review
Forex Analyst & Senior Trader
Former FX desk trader with 8 years of experience in forex and crypto markets. Expert in multi-timeframe analysis, institutional order flow, and macroeconomic fundamentals.