Gartley Pattern Trading Rules

In April 2026 I found what I was sure was a perfect bullish Gartley on GBP/JPY. X to A impulse down, B at a 61.8% retracement, C folding back, D landing right at the 78.6% of XA. Textbook. I went long at D with a $1,200 risk — 1.2% of my $100,000 account — and watched price slice straight through D like it was not there. Stop hit in forty minutes. When I measured again, calmer, D had actually printed at 70.4% of XA, not 78.6%. I had rounded. I had traded an "almost Gartley," and the market charged me $1,850 for my generosity with the ratios. That trade taught me the only thing that matters about Gartley pattern trading rules: the numbers are not suggestions.

Gartley pattern trading rules are brutally specific — B must retrace 61.8% of XA, D must terminate at 78.6% of XA, and the BC projection must land between 1.13 and 1.618 of AB. Miss any one of those and you are not trading a Gartley; you are trading a shape that reminds you of a Gartley. I now run every candidate pattern through the Candlestick Pattern Alert before I even consider an entry, because the indicator measures the legs to the decimal while I am still squinting at the screen hoping.

The "almost Gartley" that ate $1,850

Let me be precise about my mistake, because "almost" patterns are the number-one way traders lose money on harmonics. My GBP/JPY setup had X at 198.40 and A at 196.10. B retraced to 197.52 — that is 61.7% of the XA leg, genuinely fine. C dropped to 196.55. D needed to land at 78.6% of XA: 198.40 minus 0.786 × 2.30 = 196.59. Price bottomed at 196.78. That is 70.4%. Off by 19 pips on a pair that moves 19 pips while you blink.

I told myself 70% was "close enough to 78.6% in a volatile pair." It is not close enough. The 78.6% level is where the harmonic edge lives — it is the specific retracement where the pattern's historical win rate holds. Move D to 70% and you are trading a random pullback with a fancy name. I also compounded the error the way losing traders do: when price broke below my entry by 25 pips, I added to the position instead of honoring the stop, because "the pattern was still valid." The pattern was never valid. I paid $1,850 to learn that hope is not a Fibonacci level.

The deeper mistake was procedural. I had eyeballed the ratios instead of measuring them. Now I have a hard rule: no harmonic trade gets planned until every leg is measured and logged. The Candlestick Pattern Alert does the measuring for me now, and I cross-check its two most common rejections — patterns where B overshoots 61.8% and patterns where D undershoots 78.6% — against my own ruler before I risk a dollar.

Gartley ratios, decoded without the geometry headache

Forget the butterfly diagrams for a minute. A Gartley is four numbers, and if you memorize the four numbers you can validate any pattern in under a minute.

The number most traders get wrong is B. Gartley's original rule demands B retrace exactly 61.8% of XA — not 50%, not 70%. If B only reaches 50%, you are probably looking at a different harmonic (a Bat, maybe), and trading it with Gartley targets will misplace your entries and stops. Investopedia's Gartley pattern reference lays out the same ratio structure if you want a second source to memorize from.

One practical note: allow a tolerance band of about ±2% on each ratio for spread and wick noise, but never more. My GBP/JPY D was off by 8.2 percentage points. That is not tolerance; that is a different trade. If you want a second walkthrough of the ratios before you risk money on them, BabyPips covers harmonic structures in their free forex school.

Programming the Candlestick Pattern Alert for Gartley legs

I use the Candlestick Pattern Alert as a bouncer: it checks IDs at the door, and only ratio-valid patterns get in. Here is exactly how I configure it for Gartley hunting on H4 charts.

The min-leg-size filter is the unsung hero. Before I set it, the alert fired on every 20-pip wiggle that vaguely resembled a Gartley on M15, and I wasted two weeks chasing patterns with 15-pip targets and 8-pip spreads. Below 80 pips on H4, the pattern's edge drowns in transaction costs. Set the floor, and the alert goes quiet — which is exactly what you want from a filter.

How the 0.786 retracement actually behaves in live markets

Textbooks draw D as a clean bounce off 78.6%. Live markets are ruder. In my journal of 40 completed Gartleys, price touched the 78.6% zone and reversed cleanly about half the time; the other half, it overshot by 10–30 pips first — a liquidity grab below D — before reversing. The pattern still worked, but entries placed exactly at 78.6% with tight stops got wicked out.

This changes the mechanics of the entry. I no longer place a limit order at D and walk away. I wait for price to enter the D zone, then require a reversal confirmation: a bullish engulfing or a break of the M15 micro-trendline on the bullish version. It costs me 10–20 pips of the move, and it has cut my stopped-out-valid-patterns from 9 out of 40 to 3. The pattern gives you the zone; price action gives you the timing.

Stops go beyond X, not beyond D. This is a rule, not a preference: if price takes out X, the harmonic structure is broken and the pattern is dead. A stop at D-minus-a-few-pips feels cheaper, but D-zone overshoots will harvest it. My standard stop is 1.13 × the D-to-X distance beyond X, and my first target is the 38.2% retracement of the AD leg, second target 61.8%. Risk-reward on valid Gartleys averages 1:2.1 in my data, which is why the strict ratio filter matters so much — sloppy patterns dilute that ratio below profitability.

For broader harmonic context, the ABCD pattern guide covers the simpler harmonic structure that often nests inside Gartley legs, and the Wolfe Wave guide is worth reading once you are comfortable with measured moves.

Tested: 40 Gartley setups through the pattern alert

From May to September 2026 I logged every Gartley the alert flagged on H4 across eight majors, then graded them after the fact: ratio-valid or not, and what happened next. I did not trade all of them — about half were paper-traded — because the point was measuring the filter, not my P&L.

The numbers are blunt: strict patterns hit target one 65% of the time; "almost" patterns barely clear a coin flip. My GBP/JPY loser sits in that bottom row. The alert's ±2% tolerance is not being picky — it is the entire edge. Every "generous" pattern I traded before this test donated money to the market.

$100,000 account case study: one month of Gartley-only trading

July 2026. I gave myself a constraint for the month: only strict-ratio Gartleys, only H4, only with reversal confirmation at D. Nine patterns qualified across the month. Here is the full ledger — wins, losses, and the one I skipped that still bothers me.

Net for the month: +$4,300 on $100,000 — 4.3% from six traded patterns. The skipped USD/CHF pattern is the interesting row: no reversal confirmation appeared at D, so I passed, and price went on to take out X two days later. The confirmation rule saved me $1,000 that week. The lesson of the month was not that Gartleys print money — 4.3% is solid, not spectacular. The lesson was that the rules are the strategy. Every deviation I had ever made from the ratios had a negative expectancy, and a month of obedience proved it. The head and shoulders guide pairs well with this if you want a second reversal pattern with equally strict validation rules.

Gartley pattern questions, answered straight

What are the exact Gartley pattern trading rules?

B must retrace 61.8% of XA, D must terminate at 78.6% of XA, the BC projection should fall between 1.13 and 1.618 of AB, and CD typically extends 1.27 to 1.618 of BC. Stops go beyond X, first target at the 38.2% retracement of AD, second at 61.8%. Allow about ±2% tolerance for spread noise — more than that and it is not a Gartley.

How is a Gartley different from a Bat or Butterfly pattern?

The D-point retracement is the fingerprint: Gartley completes at 78.6% of XA, the Bat at 88.6%, and the Butterfly beyond 100% (typically 127–161.8%). B also differs — the Bat allows B up to 61.8% but no deeper, while the Gartley requires almost exactly 61.8%. Mixing them up misplaces your entry and stop, so identify the pattern before trading it.

Can I trade Gartley patterns on lower timeframes?

You can, but transaction costs eat the edge. Below H1, the average Gartley target is so small that spread and commission consume 15–30% of the expected move. I restrict Gartleys to H4 and D1 with a minimum 80-pip XA leg. If you must go lower, demand at least a 1:2.5 reward-to-risk after costs, not before.

What is the win rate of the Gartley pattern?

In my 40-pattern sample from 2026, strict-ratio Gartleys reached the first target 65% of the time. Published backtests vary widely (50–70%) because most do not enforce the ratio tolerances strictly. Treat any win-rate claim skeptically unless the tester states their B and D tolerances — "almost" patterns in my data only hit 33–36%.

Where should I place my stop loss on a Gartley?

Beyond X — always. If price exceeds X, the harmonic structure is invalidated and the pattern thesis is dead. I place stops at 1.13 times the D-to-X distance past X to survive spread spikes. Never place the stop just past D; liquidity grabs routinely overshoot the 78.6% zone by 10–30 pips before reversing.

Do harmonic patterns work in ranging or trending markets?

Gartleys are reversal patterns, so they perform best when completing into support/resistance in a range or at the end of a trend leg. In a strong one-directional trend, D zones get overrun and X breaks frequently. I check the higher timeframe first: if D1 is in a clean trend with no opposing structure near D, I skip the pattern.