Harami Cross — A Doji Inside the Mother Candle (Both Variants)

Harami cross looks like a beefed-up version of the regular harami: instead of a small directional candle in the middle, you get a full doji — the purest candle of indecision there is. Intuition suggests that should make it a stronger warning of a trend change than a plain harami. Thomas Bulkowski ran both variants through millions of daily stock candles, and the result came out backwards: the bullish harami cross continues the decline 55% of the time (performance rank 50th out of 103 patterns), and the bearish harami cross continues the rally 57% of the time (rank 80th out of 103 — one of the worse results in the entire catalog). Both variants statistically do the opposite of what their name promises. It's one of the cleanest examples in this whole series that the more "logical" a candlestick theory sounds, the more it deserves to be checked against the numbers before it makes it into a strategy.

What the Pattern Looks Like

Harami Cross comes in two mirror-image variants, both built from a large directional candle with a doji tucked inside it:

Bullish Harami Cross:

• Trend before the pattern: down. Without a prior decline there's nothing to reverse — a doji inside a flat consolidation is just noise, not a pattern.
• Candle 1: a tall black candle. Sellers are clearly in control of the session, the body should be long and unambiguous.
• Candle 2: a doji. Open and close sit in practically the same place. The doji's entire range — including its wicks — fits inside candle 1's price range.

Bearish Harami Cross: the mirror image — the trend before the pattern is up, candle 1 is a tall white candle, candle 2 is a doji that fits (wicks included) inside its range.

The difference from a regular harami looks cosmetic at first glance: there, the second candle has a small but visible directional body; here, that body all but disappears. The textbook logic says that since a doji is the purest picture of market indecision, harami cross should be a stronger stall-and-reverse signal than a plain harami. Bulkowski himself calls this reasoning the "theory" and immediately sets it against the "reality" from his testing — and for harami cross, those two things diverge unusually hard.

On crypto the pattern needs no adjustment to its definition. There's no gap involved here — unlike, say, the rising/falling window or the kicker pattern — so the "large candle plus doji inside it" geometry works identically on a 24/7 market as it does on stocks with a trading session. On BTC/ETH on the D1 and H4 timeframes, harami cross shows up regularly, especially right after a sharp directional move, when the market suddenly freezes before its next decision.

What the Numbers Say

Results from Bulkowski's tests (hundreds of measured "perfect" trades on US stock daily data), for each variant separately:

Bullish Harami Cross:

• Theory: bullish reversal. Data: continues the decline 55% of the time. That's two percentage points worse (for the theory) than the plain bullish harami (53% reversal success) — harami cross doesn't just fail to reinforce the signal, it tips the balance even further toward continuation.
• Overall performance rank: 50/103. Dead center of the pack of 103 candlestick patterns — the move after the breakout is average, unremarkable in either direction.
• Frequency rank: 47/103. The pattern shows up regularly; it isn't a statistical rarity like, say, the kicker pattern.
• Best average 10-day move: 4.52% (bear market, upside breakout), with that result ranking 36th out of 103. Bulkowski considers 6% or more a good result — the pattern doesn't reach that bar even in its best of four market/direction combinations.
• Best price-target completion rate: 74% (bull market, upside breakout) — the one genuinely strong number in the whole set, but it measures how far the move travels, not which direction the breakout goes.

Bearish Harami Cross:

• Theory: bearish reversal. Data: continues the rally 57% of the time. That's better than the 53% continuation rate of the plain bearish harami — but "better" here means "works even more strongly against its own name," not "reverses the trend more reliably."
• Overall performance rank: 80/103. Bulkowski calls this outright "downright awful" — it sits in the bottom fifth of the catalog, worse than the vast majority of candlestick patterns, including its own bullish mirror image.
• Frequency rank: 45/103. About as common as the bullish variant.
• Best average 10-day move: -3.13% (bear market, downside breakout), while the best rank for that move (41/103) actually came from a different combination — a downside breakout in a bull market. Bulkowski doesn't hide his puzzlement: the pattern's best results come from moves AGAINST the prevailing market trend, which he chalks up to a snapback rally or drop rather than to the pattern itself.
• Best price-target completion rate: 69% (bull market, upside breakout).

Three extra tidbits from the data apply to both variants: patterns appearing within a third of the yearly low or high perform best; pick tall first-day candles; and for the bearish variant, the best reversal results come from patterns that appear after an earlier upward retracement of a downward trend — meaning inside a counter-trend bounce, not inside a clean uptrend.

The standard disclaimer applies at full strength here: the data comes from US stocks on the daily timeframe. On crypto we carry over the logic and geometry of the pattern, not the exact percentages — a different 24/7 volatility and liquidity regime almost certainly shifts these numbers, though the direction of the takeaway (the theory-vs-data gap) probably doesn't flip.

How to Trade It (and How Not To)

How NOT to trade it: treat a doji inside the mother candle as a reliable reversal signal just because it's "the purest indecision candle there is." The data contradicts that in both directions at once: the bullish variant more often continues the decline, the bearish variant more often continues the rally. Entering "on the harami cross" without confirmation is a bet where the statistics work against you, not for you.

Scenario 1 — treat harami cross as a flag that says "the market hesitated," not as a directional signal. A doji inside a large candle's body really tells you one thing: the previous candle's momentum evaporated, and market participants froze. That's information about the market's state, not a forecast. Make your decision based on what happens AFTER the pattern — a close outside its range — not on the pattern's mere appearance.

Scenario 2 — if you trade it directionally, demand confirmation and level context. The preferred setup from the data: a pattern appearing within a third of the yearly extreme (the low for the bullish variant, the high for the bearish one), built from tall candles. Enter only after a close beyond the pattern's extreme in the direction you actually want to trade — not the direction you assumed going in, since the pattern more often runs the other way than its name suggests.

Scenario 3 — for the bearish variant, consider trading the continuation instead of the reversal. Since the data says the market keeps rising 57% of the time after this pattern, it may make more sense to wait for confirmation of trend continuation (a close above the pattern's high) than to try catching a reversal down. That's the exact opposite of what the textbook teaches — and exactly what 4.7 million candles show.

Stop loss and target. Place the stop outside the entire pattern's range — below the low for a long, above the high for a short — counting both candles' wicks, not just their bodies. Target: the nearest real structural level, with the humility the data demands — the best average 10-day moves (4.52% and -3.13%) don't even reach the 6% threshold Bulkowski calls "good." Size your position from the stop, not from how convincing the signal feels — with this pattern, conviction and data diverge unusually sharply.

Quick checklist:

• Is the second candle a genuine doji (open ≈ close), not a small directional candle — the trait that separates harami cross from a plain harami?
• Does the doji's entire range, wicks included, fit inside the first candle's price range?
• Does the pattern sit near a yearly extreme, not in the middle of a random consolidation?
• Do you remember which variant statistically continues rather than reverses (bullish → decline, bearish → rally)?
• Are you entering only after confirmation — a close beyond the pattern's range — instead of on the doji itself?
• Is the stop outside the whole pattern including wicks, with a modest target that matches the data, not the pattern's name?

Myth vs Measurement

Myth: "A doji inside the mother candle is a stronger reversal signal than a small candle in a regular harami." Measurement: harami cross underperforms the plain harami on both sides. The plain bullish harami reverses 53% of the time (rank 38/103); the bullish cross reverses only 45% of the time (i.e., continues 55%, rank 50/103). The plain bearish harami reverses 47% of the time (53% continuation); the bearish cross reverses just 43% of the time (57% continuation, rank as low as 80/103). The doji doesn't strengthen the signal — it weakens it.

Myth: "The pattern's name tells you which way to trade." Measurement: both labels ("bullish reversal," "bearish reversal") describe the direction that statistically occurs less often than its opposite. This isn't a pattern with no edge — it's a pattern whose modest but real edge sits on the opposite side from what intuition suggests.

Myth: "Bearish harami cross is a solid sell signal at the top." Measurement: an overall rank of 80 out of 103 is one of the worse results in the entire candlestick catalog — worse than the vast majority of patterns, including its own bullish counterpart. The name sounds ominous; the data says otherwise.

Myth: "Since harami cross is a beefed-up harami, it should behave more decisively in both directions." Measurement: it does behave more decisively — just in the direction of continuation, not reversal. That's the exact opposite of the reinforcement candlestick theory assumes, and exactly why this pattern is a good exercise in reading data instead of names.

Harami cross is a pattern where the legend and the measurement don't just diverge — they point in opposite directions. A doji inside the mother candle's body looks dramatic on a chart and sounds convincing in a description: "extreme indecision after a strong move must signal a turn." Four point seven million candles say otherwise: the bullish variant more often drags the market lower, the bearish variant more often drags it higher, and the overall quality of the move after the breakout sits in the bottom half of the 103-pattern field in both cases. Anyone who only knows the name trades a reversal that statistically isn't there. Anyone who knows the numbers checks the data first — and only then decides which direction is even worth watching.

FAQ

Rafał — Krypto Bez Ściemy

Trader and founder of Strefa Tradingu. He’s been breaking crypto down on YouTube for years — no hype, no signals, with a focus on market structure and risk management.

kryptobezsciemy.plInstagramYouTubeKlub Zero