Gartley pattern — the original harmonic formation at 0.786 XA

Gartley describes five ordered turning points X, A, B, C and D under a selected convention. Check all ratios using the same anchors and record when the points become known. A projected D zone is distinct from a confirmed turn, an available entry and an actual fill. One 0.786 level alone does not identify a complete pattern.

What the Gartley pattern is and where it came from

Carney’s public account traces the name “Gartley 222” to page 222 of H.M. Gartley’s 1935 book, Profits in the Stock Market. He distinguishes that original framework from later specific Fibonacci retracements, attributing his own version to The Harmonic Trader. This is the source author’s historical attribution, not evidence that every harmonic method began with this one or that its ratios produce the most reliable reversals.

Carney specifies B= 0.618 XA and D= 0.786 XA, with an equivalent AB=CD and BC projection 1.27 or 1.618. Volume One allows C from 0.382 to 0.886 AB. TradingView’s tool description instead gives BC/AB 0.618–0.786 and CD/BC 1.27–1.618 as ranges. A range is not a test of its endpoints only. The basics of trading harmonic patterns explain why the definition, denominators and tolerance must be named beforehand.

Structure and Fibonacci levels

Gartley ratios identify a candidate completion area. Confirmation, invalidation, costs and position size are separate parts of the trading plan.

Use absolute price distances on a linear scale. AB/XA measures B, AD/XA measures D from A toward X, BC/AB measures C and CD/BC measures the final projection. AB=CD here means equal price-leg distances, not equal durations or equal diagonal line lengths on the screen. A different B fails the selected condition but does not itself establish a different complete harmonic pattern.

For our comparison exercise, choose B and D at Carney’s references, AB=CD, and the two continuous ratio ranges stated in TradingView’s Gartley description. This is an explicit operational combination, not a claim that every scanner or the narrower endpoint-based author definition accepts it. Use Fibonacci extensions when projecting targets to calculate the chosen relationships without adopting promotional success claims.

XA = 200 pips; AB and CD each 123.6 pips; BC = 90 pips; AD = 157.2 pips. Construct B=A−0.618(A−X), D=A−0.786(A−X), then C=D+(A−B) for equal legs. CD/BC is the reciprocal of BC/AB under exact AB=CD, not independent evidence of the future. Our 1.373333 passes a continuous 1.27–1.618 range, but is not equal to 1.27 or 1.618. An endpoint-based test can reject it. The original rounded B 1.0876 would give 0.620, not exact 0.618; avoid labels inconsistent with the actual supplied points.

How to recognise the formation step by step

Step 1 — find the X-A leg and the trend context

Choose the series, interval, linear scale and pivot-confirmation rule before measuring. A higher-timeframe trend filter is an additional hypothesis, not automatically the most reliable direction. Record when each anchor became available; a retrospective turning-point label cannot justify an earlier order.

Step 2 — confirm point B at 0.618 of the XA leg

Test B against 0.618 XA using a tolerance set in advance. In our exercise ratio tolerance ±0.005 is an absolute 0.5-percentage-point deviation, not 0.5% of the target coefficient. At XA 200 pips it permits 1 pip of AB-distance error. A shallower B does not establish Bat without its D and remaining conditions.

Step 3 — check the B-C leg and the completion of point D

Check all remaining ratios and structural order together. Our illustrative tolerances for B/D are ±0.005 and for CD/AB ±0.020, with inclusive continuous BC bands. These are exercise inputs, not optimal or universal author settings. Changing a tolerance after seeing a reversal changes the test rather than validates an original candidate.

Here X=1.08000 and A=1.08200; a 0.005 ratio tolerance allows only 0.1 pip. Changing display precision can alter qualification if rounded numbers are fed back into the test. Calculate from the actual available source prices before display rounding. If the source really has only four decimals, do not invent missing digits. This comparison checks two conditions only, not a complete Gartley or a required minimum XA length.

Entry, stop and targets — a hypothetical example

Projected D can be calculated from earlier anchors, but an actual turning extreme may need later bars to confirm. Keep times for projection, touch, confirmation and any fill separately. A hammer or engulfing filter is another defined condition to evaluate, not proof that entering at the prior minimum was possible. Crossing X violates our chosen inside-X geometry, but does not supply a universal stop distance or identify another pattern automatically.

If separately choosing target references at 0.382 and 0.618 AD from D, our AD 157.2 pips gives 1.09028504 and 1.09399496. Reference distances from D are 60.0504 and 97.1496 pips. A chosen stop reference 1.07980 is 44.8 pips below D, so raw distance ratios are about 1.3404 and 2.1685 before fills and costs. These are conditional arithmetic, not actual executable entries, inherently conservative targets or expected profits. A delayed entry, different exit sizes and costs change reward:risk; it does not depend only on XD.

Define an invalidation level and planned stop before entry; this example uses a stop beyond X. An ordinary stop does not guarantee a ceiling on actual loss because execution can slip. Swing-point selection is also discretionary, so record the chosen X, A, B and C points and size the trade using the actual entry-to-stop distance.

The most common mistakes when trading the Gartley

1. Accepting B/D labels after rounding without checking the supplied prices, denominator and stated tolerance.
2. Assigning an earlier fill from a turning extreme or confirmation known only later.
3. Treating a projected geometric zone as proof of an entry or future reversal.
4. Relabelling a failed Gartley as Bat based only on a deeper D without checking its B and remaining conditions.
5. Changing the pivot, trend filter, continuous versus named-level definition or tolerance after observing the outcome.

How the Gartley differs from other harmonic patterns

The Bat pattern references D to 0.886 XA and requires B below 0.618, with 0.382/0.500 preferred. The Gartley and Bat patterns guide compares the full conditions. Carney’s Ideal butterfly pattern instead uses D 1.27 XA beyond X; regular crab pattern uses D 1.618 XA. Named variants differ. These references separate retracement and extension constructions, not a ranking of gentleness, stop tightness or profitability.

Who this pattern is for

The technical analysis material and Fibonacci retracements explain measurement. A fixed formula is reproducible once anchors and scale are stated, while discretionary choices need explicit records. Oscillator or support filters are additional hypotheses, not automatically the best combination. Retain rejected and failed cases and compare unchanged rules on separate later data.

What to do tomorrow to start learning the Gartley pattern

1. Select a series and pivot-confirmation rule, recording when each XABCD point became known. Keep price distances separate from screen-line length or equal time spacing.
2. Record all raw ratios, named author/tool convention, continuous versus endpoint tests and tolerances before inspecting later outcomes.
3. Keep projection, first touch, confirmed turn and any execution in separate fields. Retain failures, costs and missed fills as well as successful examples.
4. Use an alert to locate a candidate for checking. It does not establish a complete pattern, independent confirmation or an executable fill at D.
5. Evaluate frozen entry/exit rules on later data; twenty demo trades or a repeatable hit rate alone establishes neither future profitability nor live readiness.