Gartley and Bat — ratios, candidate D prices and execution risk

Gartley and Bat share X–A–B–C–D geometry, but their names are not interchangeable just because D is near an expected price. Compare AB first, then the remaining legs and projection convergence. This lesson uses Carney’s conventions as definitions and hypothetical arithmetic, not evidence of investment performance.

The shared skeleton of five points

Five ordered turning points define four legs: XA, AB, BC and CD. Use positive leg lengths and name each ratio’s denominator. In a bullish model with X below A, B = A − rB × (A−X), and the XA-based candidate D = A − rD × (A−X). In the bearish mirror, X is above A and the level is A + r × (X−A). Do not multiply the price A itself by the ratio or measure this retracement from X.

That is where the characteristic outline comes from, the one the eye picks up first. A bullish pattern, ending in a low at point D, draws the letter M on the chart: two peaks at points A and C, with three lows between and beneath them. A bearish pattern, ending in a peak at D, gives the mirror image, a W. Symmetry alone, however, settles nothing — two setups that look almost identical may in fact be different schemas and demand a different trade plan.

What actually separates a Gartley from a Bat

In Carney’s convention, Gartley has AB/XA=0.618 and AD/XA=0.786, with converging AB=CD and a BC projection of 1.27 or 1.618. Bat uses AD/XA=0.886, B below 0.618 with 0.382 or 0.50 preferred, a BC projection of at least 1.618 and at least an equivalent AB=CD, with 1.27 extension preferred. A narrow 0.382–0.50 B filter is a test choice, not the entire Bat description. The harmonic pattern trading guide expands the measurement process.

How to identify each pattern step by step

Use one recorded pivot-selection rule and tolerance. For a common hypothetical EUR/USD frame X=1.1000, A=1.1200, XA is 0.0200 or 200 pips at 0.0001. Gartley candidate B=1.10764 and XA-based D=1.10428; choosing B=0.50 XA for Bat gives B=1.11000 and D=1.10228. D is respectively 42.8 and 22.8 pips above X, a 20-pip difference. These are candidate coordinates only: C, complete candles and an entry condition are not supplied.

Adding C=1.11500 to the Gartley candidate gives AB=0.01236, BC=0.00736 and CD=0.01072. An exact AB=CD projection gives D=1.10264, 16.4 pips below the XA-based 1.10428. It fails our illustrative maximum 2-pip difference between these projections; that tolerance is our rule, not Carney’s. With C=1.11650 instead, the projected D=1.10414 differs by 1.4 pips and passes only this one convergence check. The remaining geometry and trade conditions still require assessment.

For a bullish XA of length R, let rB=AB/R and rC=BC/AB. Exact AB=CD implies AD/R = rB × (2−rC). With rB=0.618 and rC=0.382 the result is 0.999924, not 0.786. To also impose AD/R=0.786 requires rC = 2−0.786/0.618 ≈ 0.728155. The ratios cannot be independently assigned popular values without checking coordinate consistency.

Entry at point D, stop and targets

The potential reversal zone does not set an entry or a loss ceiling. Separate geometry rejection, entry confirmation and the protective order. Waiting for a candle close cannot justify a retrospective fill at ideal D. A stop beyond D and an instruction to exit only after X is exceeded describe different rules; define which is active and record its price side and execution.

Targets at 0.382 or 0.618 of an identified AD leg are possible exit rules, not distances that a rebound usually covers. For a bullish model calculate them upwards from D; for a bearish model downwards from D. Compare a full exit with any split exit under the same data, fees and fills. The Fibonacci retracements guide explains the measurement convention; it does not establish which exit is superior.

A hypothetical example: bearish Bat on EUR/USD

Keep a separate bearish EUR/USD arithmetic example: X=1.0950, A=1.0750, B=1.0850 and C=1.0790. XA=200 pips, AB/XA=0.50 and BC/AB=0.60. The exact XA-based D is 1.0750 + 0.886 × 0.0200 = 1.09272, a 177.2-pip AD. Here CD/BC ≈ 2.2867 and CD/AB=1.372. These calculations alone do not certify a complete Bat or justify an entry; record convergence tolerances and all required checks.

Independently assume an actual Bid short fill at 1.09272, an Ask stop fill at 1.09600, and a pip of 0.0001. The planned price distance is 32.8 pips; the stop is 10 pips above X. Chosen AD exit fractions give Ask 1.08595096 at 0.382 and Ask 1.08176904 at 0.618, gains of 67.6904 and 109.5096 pips under these assumed fills. The second reward/planned-risk ratio is 109.5096/32.8 ≈ 3.3387 before commission or funding. These Ask prices are hypothetical execution assumptions, not guaranteed fills at chart levels. Crossing X and triggering the stop at 1.09600 are different events.

Compare two other independent hypothetical buys using the earlier bullish candidate D prices plus 2 pips: Ask 1.10448 and Ask 1.10248. Both use 10,000 EUR, a USD account, an ordinary Bid stop at 1.09900 and total commission USD 2, excluding funding and conversion. The distances are 54.8 and 34.8 pips; exact stop execution gives net losses USD 56.80 and USD 36.80. A worse Bid fill at 1.09800 gives losses USD 66.80 and USD 46.80. Use 10,000 × (Bid exit − Ask entry) − 2; spread is already in the prices. This difference follows from chosen entry prices and equal volume, not greater safety of Bat, and none is a maximum possible loss.

The limits of the method and the honest caveats

Record the full coordinates, data precision, pivot rule, tolerance, recognition time and result for every candidate. A pivot needing later bars is not knowable at its own bar. Additional support, averages or timeframe filters are separate rule variants; several conditions derived from the same prices do not automatically constitute independent evidence of a reversal. Test the rules on a different period without silently refitting them.

Check for inconsistent ratio denominators, altered anchors, retrospective fills and omitted costs. Report rejected patterns, no entry and losses as well as profitable examples. A shape label cannot establish the net result of a strategy.

What to do tomorrow

1. Choose a fixed dataset and pivot rule. Record full XABCD coordinates and the time each point becomes knowable; keep rejected candidates rather than selecting only successful reversals.
2. Write the selected definition and your own tolerances separately. Check each ratio and projection convergence; exact B and D do not validate the remaining legs.
3. Define an entry condition, its expiry and geometric cancellation before testing. Treat a candle or higher-timeframe filter as another specified variant, not a method that automatically eliminates most false signals.
4. Compare full-position and partial exits as separate simulations, including fees, funding and assumed fills. Calculate potential reward and planned risk without treating their ratio as a success probability.
5. Keep a journal of candidate coordinates, available information, classification, entry or no-entry decision and costs. A chosen month can be a practice period; its observations are not automatically enough to demonstrate effectiveness.

Related reading: Fibonacci retracements are the foundation on which both patterns rest; the Gartley pattern strategy develops a practical trade plan; the Bat pattern strategy shows how to use the deeper point D, and harmonic pattern trading places the topic in the wider XABCD context. For broader background see the technical analysis section on ForexMechanics.com.