Three Black Crows: The Exact Rule and Real Examples
Three black crows is distribution written in plain handwriting. Sellers take a session, hand back a little of it at the next open, and take it again — three times, with each red body starting inside the one before it and finishing lower than that one did.
Every guide describes the staircase. Almost none say which comparison decides it, and fewer still have counted. Vike's batch_three_black_crows flagged 79 instances across 479,749 daily candles from 376 Binance pairs, more than double what its bullish mirror managed on the same bars, and 43 of the 53 old enough to grade kept falling. Here is the arithmetic, five of those detections, and one that rallied 48% instead.
What three black crows actually means
Supply arriving in instalments, not in a panic. Each session opens back up inside the previous red body, which hands buyers a starting price better than yesterday's close, and each session still ends lower than the last one. Three attempts, three rejections.
Compare that with a single vertical red bar. One capitulation candle can be a cascade of forced liquidations that finishes as soon as the margin calls do. A staircase says something slower and worse for holders: sellers were patient enough to keep working the same direction across three separate sessions, and every bid they were offered got used up.
Context governs how much of it to believe. Following a stretched rally the shape is a topping signal with real weight. Halfway into an established downtrend the same three bars are just Tuesday, and I've been filled at the low of the week doing exactly that.
How to identify three black crows: the exact rules
batch_three_black_crows measures. Dashed lines are the containment comparisons, the bracket is the lower-shadow limit, and the wicks above the bodies are drawn long on purpose: the function loads the low column and never the high.Five conditions, evaluated in this order, and every one of them arithmetic on the open, close and one shadow.
- All three bars close below their open.
- Each body is at least 0.7 times the 10-bar average body.
- Each bar opens inside the previous bar's body — below that bar's open, above its close.
- Each lower shadow is no more than 30% of its own body.
- The three closes fall in sequence.
How Vike computes it.batch_three_black_crowsreads three bars at a time and stamps the last of them-100when all five conditions hold. The 10-bar average it sizes bodies against already includes those three candles, so a violent slide raises the threshold that slide has to clear.
That third condition is stricter than it looks in prose. A market with no session break usually opens precisely where it last traded, and precise equality fails a strict comparison. Across 58,402 red triples in this tape, 34,308 of them — 58.7% — were eliminated by an exact tie before any other clause was consulted.
The clause that does the work
Ask a trading forum what makes three black crows hard to find and you will hear about the staircase. The measurement points somewhere else entirely.
From those 58,402 red triples, 8,285 carried three bodies of sufficient size. Demanding that each bar open back inside the previous body left 939. Then the lower-shadow test cut 939 to 79 — it discarded 91.6% of everything still standing, and hardly any summary of this pattern mentions shadows at all.
A second detail is stranger. Having survived the shadow test, a candidate still has to show three descending closes, and that final clause removed zero candidates in half a million bars: 79 went in, 79 came out. Logically it can bind, and you can draw a triple where it does. Empirically, on this tape, it never did.
Read the two facts together and the rule collapses into something you can hold in your head. Bodies of real size, each opening back inside the last, each closing near its low — with the closing-near-the-low part quietly doing nine tenths of the filtering.
What the rule never checks
Upper shadows are invisible here. batch_three_black_crows loads the open, low and close columns and never reads the high, so a crow can carry a rejection wick of any height above it and remain a perfectly valid crow. Its bullish mirror is blind on the opposite side.
No trend is required either, nor a level, nor volume, nor any relationship between the three body sizes. Three candles and their 10-bar context are the entire input.
Which explains this pattern's least convenient habit: it is largely a detector of market-wide events. Ten of the 79 detections landed on a single day, 2022-06-12 — BCH, COMP, DOT, FIL, IOTX, LPT, NEAR, SHIB, TFUEL and VET all printed it during the Celsius weekend. That was not ten independent confirmations of anything. It was one deleveraging, sampled ten times.
Three black crows vs three white soldiers
Vike's crate writes the two directions as two functions, mirrored clause for clause: red becomes green, the containment comparison flips which end of the body it means, descending closes become ascending ones. Neither one is a sign flip of the other, which is how bearish engulfing relates to its bullish twin — that pair shares a single function and splits on the sign alone.
The mirror also swaps the price column being loaded, and that is the part worth memorising. Crows read the low and police the lower shadow; three white soldiers read the high and police the upper one. Whatever hangs off the other side of the candle is simply not part of the calculation.
Outcomes diverged sharply too. Crows continued in 43 of 53 graded cases against 14 of 25 for soldiers, though I would not read that as the bearish pattern being cleverer. Crypto falls faster than it rises, so a fixed 4% follow-through threshold is easier to hit on the way down, and part of that gap is the threshold rather than the shape.
How to trade three black crows
Use it as a location and a warning, never as a market order on the third close.
Sell the bounce, not the third bar. By the time crow three closes, the slide has already covered 19.3% at the median across every detection, and the extreme was 46.4%. Chasing at that point puts your stop three sessions above your entry.
Put the invalidation above the pattern high. The highest print across the three bars is the level that says sellers lost the argument. It will usually be a long way up, and cutting the stop tighter to feel comfortable converts a decent idea into a random one.
Watch the third body. When the last crow is the biggest of the three, I treat it as more likely to be exhaustion than continuation, and I have been stopped out often enough to keep treating it that way.
Check the calendar before the chart. If ten names on your screen print crows at once, that is a macro event and your edge is in position sizing, not in the pattern. A quick RSI reading below 30 across the board tells you the same thing faster.
Where three black crows fail
BEL on 2021-07-20 is the one I keep for perspective. Three qualifying red bodies, lower shadows of 6%, 9% and 27%, a 29.5% slide across the sequence — every clause satisfied, nothing marginal about it. Price then declined a further 2.5% and rallied 48% inside five sessions.
That instance is the argument for confirmation in one chart. My first year of trading was full of shorts entered on exactly that bar, and the reason is easy to state afterwards: a detection measures three sessions that already happened, and a market down 29.5% in three days is by definition closer to oversold than it was on Friday.
Fakeouts of this shape cluster in exactly the places you would guess: at the end of forced-selling episodes, in thin books after a listing pump unwinds, and directly into a level that has already absorbed sellers once. I learned the last one shorting a third crow straight into a range floor that had held four separate times, which cost about as much as a lesson can.
Frequently asked questions
Is three black crows a bearish reversal or a continuation? Both, depending only on where it sits. After a rally it argues that supply has taken control; inside a downtrend it merely restates the trend. The detector itself does not know which case it is in, since it reads no history beyond a 10-bar body average.
How reliable is three black crows? Better than its bullish mirror on this data — 43 of 53 graded detections kept falling — but it fired once in roughly 6,000 bars, so it is far too rare to build a system around and far too correlated across symbols to count as independent evidence.
Do the wicks matter? Only the lower ones, and they matter enormously: each must stay within 30% of its own body, a test that removed 91.6% of otherwise-qualifying triples. Upper wicks are never inspected.
What is identical three crows? A sibling in the same source file, and not a stricter version of this one. It keeps the colours, the body sizes and the descending closes, then drops the shadow test altogether and asks instead that each open land within 5% of an average body of the previous close. Note what that permits: an exact tie, the very thing a strict containment comparison rejects. One rule is built around the equality that kills the other.
Can three black crows appear at a bottom? Yes, and that is its most expensive failure. The BEL example above is precisely that case: three textbook crows printed into a level that then held, and anyone short at the third close spent the next week watching a 48% rally.
This is educational material, not financial advice. Candlestick patterns describe probabilities, not certainties, and every real trade carries real risk — size positions so a wrong read costs little.