AB=CD, Bearish

Core idea

The bearish AB=CD is a harmonic pattern: four turning points (A high, B low, C high, D high) whose locations are governed by Fibonacci ratios rather than free-form swing geometry. The headline finding from Bulkowski's tests is split: the pattern is outstanding at predicting where point D will form (price reaches the calculated D level ~95% of the time or more) and mediocre at predicting that price will reverse there (only 32% to 38% of the time). The pattern works as a price-projection tool first and a reversal signal second.

Why this asymmetry matters

A pattern that nails the target but not the turn changes the trade. Instead of fading at D — the textbook play — you can ride toward D from C, take profits at or near it, and treat any reversal there as a bonus rather than a trigger. The Fibonacci skeleton is doing real work; the "and it'll turn" claim is not.

Why it matters

Most chart patterns ask the trader to wait for a break. Harmonic patterns try to skip that wait by calculating the turn in advance. When they work, you have an early, high-probability entry. When they don't, you have a clean rule for stepping aside. Either way, you only need a computer (Patternz, Bulkowski's free tool, will find them) — manual identification is not realistic. In the bull market, the average decline after D is just 12.7% with a 26.3% breakeven failure rate, putting the bearish AB=CD at the bottom of the five bearish Fibonacci patterns Bulkowski studied.

Key takeaways

Mental model — the harmonic skeleton

Copy sourceZoom

Identification guidelines

• BC/AB retrace. Compute (B − C) / (B − A). The result must equal one of the listed Fibonacci ratios (~.382, .5, .618, .786) within tolerance (~1%).
• DC/BC extension. Compute (D − C) / (B − C). Must equal 1 / R where R is the BC/AB retrace.
• No interim violation. Between C and D, no minor low may drop below C. If one does, the pattern search terminates.
• Duration. Limit to six months (an arbitrary cap to keep patterns useful).
• Volume. Downtrending most often, but near random — do not reject a pattern for volume alone.

How the pattern fails

Two distinct failure modes:

1. Price never reaches D. A minor low forms below C en route. Rare (~5%) but invalidates the pattern entirely — there is no D to trade.
2. Price reaches D but does not reverse. Far more common (62–68%). Price continues higher past D, sometimes by a wide margin. Shorts placed at D suffer.

A third lighter failure — the "5% failure" — happens when D does form a minor high but the subsequent decline is less than 5%, not enough to cover slippage and commissions.

Practical application

A trade plan that respects the statistics

1. Wait for software to identify ABC. Compute the projected D price and date.
2. If you wish to trade long-to-D, buy on the first close above the C-pivot bar's high. Stop below C.
3. Exit at D regardless of what happens there. If D forms a minor high and the next bar closes below D's low, consider a short with a stop above D — but size small and exit on any close back above D.
4. In bear markets, prefer downward-volume-trend AB=CD patterns (24% average decline vs 19%). In bull markets, volume gives you no edge.

Example: AB=CD on a synthetic mid-cap

A fabricated MID stock prints A at 56.81, B at 63.85, C at 60.32. The BC/AB retrace is (63.85 − 60.32) / (63.85 − 56.81) = 0.501 ≈ .5 — qualifies. Predicted D = C + (B − C) / 0.5 = 60.32 + 7.06 = 67.38. Price drifts higher over 8 sessions and prints 67.36 — within 0.03 of the target — and forms a minor high.

A trader who bought at C+pivot-confirmation around 60.80 with a stop at 60.10 (just below C) and sold at the projected D of 67.30 captures ~10.6% with ~1.1% risk — about 9:1 reward-to-risk on the high-probability projection. A trader who instead waited at D to short pulls 3% (price drops to 65.22 = E) before the pattern fails and price grinds higher — a much worse trade despite using the same pattern.

Where this fits in the family

AB=CD is the simplest Fibonacci pattern. Bat (Bat, Bearish), butterfly, crab, and Gartley extend the same idea with five turns and tighter ratio constraints. They are rarer, harder to qualify, and often perform better when they do qualify. AB=CD trades the volume; the others trade the quality.

Related lessons