Harami Pattern: 243 SPY Cases Against the Base Rate
A harami is a two-candle pattern in which a small body sits entirely inside the body of the larger candle before it, with the two bodies in opposite colors. A bullish harami follows a decline and a bearish harami follows a rise, and each is read as a sign that the move before it is losing force.
The harami is the quiet cousin of the engulfing pattern. Instead of a second candle that swallows the first, the second one is small and tucked inside it, like a pause after a loud day. This page sets out an exact rule, walks through two SPY cases from 2026, and counts every case on SPY since 1993 and on five large caps since 2000.
How it forms
The first day is a large candle in the direction of the recent move. In a decline it is a long red body, in a rise a long green one. It says the side in charge pressed hard.
The second day opens back inside that body and stays there. Its open and its close both fall between the first day’s open and close, and it closes in the opposite color. The body is small, so the session went nowhere much, and the side that had been pressing did not follow through.
The name is the Japanese word for pregnancy. The long first body is read as the mother and the small second body as the child inside it. When the second day is a doji, opening and closing at almost the same price, the variant is called a harami cross.
The pattern compares bodies only. Shadows can poke out beyond the first day’s body, which is the main difference from an inside bar, where the whole high-to-low range of the second day must fit inside the first. It is the mirror of a bullish engulfing: there, the second body covers the first; here, the first covers the second.
The exact rule counted here
Bullish harami, on daily bars. Day 1 closes below its open. Day 2 closes above its open. Day 2’s open and close are both strictly above day 1’s close and strictly below day 1’s open. Day 1’s body is at least the average body of the 20 sessions before it. The close the day before day 1 is below the close 10 sessions before that, so prices were falling going in.
Bearish harami is the mirror. Day 1 green, day 2 red, day 2’s body strictly inside day 1’s, the same long-body test, and the close before day 1 above the close 10 sessions earlier.
Outcome. The close five sessions after day 2, compared with day 2’s close. Confirmed means the next session closed beyond the far edge of day 1’s body: above its open for a bullish harami, below its open for a bearish one.
Ties stay ties. In this file every SPY bar from 29 January 1993 to 26 January 2001, 2,019 of them, is priced in exact fractions of a dollar, where equal prices are common. Prices are compared unrounded, so a body edge that exactly equals the edge before it does not count as inside.
A worked example
Monday 27 July 2026 set it up. SPY opened at $744.91 and closed at $739.09, a red body of $5.82. The average body over the 20 sessions from 26 June to 24 July was $2.27, so day 1 was about 2.6 times normal. The close on 24 July, $738.93, was below the $754.95 close of 10 July, so the decline rule held.
Tuesday 28 July was the small green child. It opened at $739.19 and closed at $740.86, a body of $1.67 that sits between $739.09 and $744.91. Its high of $742.79 and low of $735.98 were also inside day 1’s range, so this one was an inside bar too.
The next day went against it first. On 29 July SPY closed at $729.46, below both candles, so the harami never confirmed. Five sessions after day 2, on 4 August, SPY closed at $771.33, 4.11% above the $740.86 close. The count below scores that as higher, and it shows why one case says little: the path ran through a lower close before the rise.
The original data
Across SPY daily bars from 29 January 1993 to 25 September 2026, 243 haramis met every rule: 127 bullish and 116 bearish. All 243 have five later sessions to measure.
Bullish haramis leaned up. 82 of the 127 closed higher five sessions later, 64.6%, with a median change of +0.83%. The comparison is every SPY day after the same kind of decline: 1,895 of 3,315 closed higher five sessions later, 57.2%. An exact binomial test puts a gap that size at p = 0.106, so a sample of 127 drawn at random from those days would show a gap at least that large, in either direction, about one time in ten. That is within chance.
Bearish haramis leaned down, by a similar margin. 56 of the 116 closed lower, 48.3%, against 2,111 of 5,123 days after a rise, 41.2% (p = 0.131, also within chance). The median change was +0.06%, so the typical case was flat rather than lower.
Five large caps gave smaller gaps on bigger samples. Across AAPL, MSFT, JPM, XOM and KO from 2000, 292 of 511 bullish haramis closed higher (57.1%, against 54.8% after any falling stretch, p = 0.286) and 264 of 546 bearish ones closed lower (48.4%, against 45.8%, p = 0.23). Every case is listed in the harami events file.
Confirmation made the SPY numbers worse, not better. 35 bullish haramis were confirmed, and 13 of them (37.1%) closed higher five sessions after the confirming day. 27 bearish ones were confirmed, and 11 (40.7%) closed lower. Both samples are small, but neither points the way the extra wait is meant to.
Harami or inside bar
The two overlap but are not the same. Without the long-body and trend rules, SPY printed 628 two-day harami shapes: 349 bullish and 279 bearish. 254 of those, 40.4%, were also strict inside bars, with the second day’s high below the first day’s high and its low above the first day’s low. The other six in ten had a high or low that reached the first day’s high or low, or went past it.
Most inside bars are not haramis. SPY had 898 strict inside bars over the same period, and only 254 of them were harami shapes; the color and body rules leave out the rest.
The filters did not change the picture much. The 349 bullish shapes closed higher five sessions later 63.0% of the time, close to the 64.6% for the filtered set, so the long body and the trend rule mainly cut the count. On the full rule, 104 of the 243 haramis were also inside bars.
When it fails
The pause becomes a continuation. On Friday 13 March 2026 SPY printed a long red day, $669.27 to $662.29, a body of $6.98 against a 20-day average of $4.18. On Monday 16 March a green body from $668.38 to $669.03 sat inside it, and on 17 March SPY closed at $670.79, above the first day’s open, so the bullish harami confirmed. Five sessions later, on 24 March, SPY closed at $653.18, 2.6% lower.
The lean is too small to stand on. The SPY gaps of about seven points could come from chance roughly one time in ten, and on the large caps the gaps shrank to under three points. A trader acting on every bullish harami was buying into a setup whose record sits close to that of any day after a decline.
Timing is loose. The five-session window is a choice. The 28 July 2026 case was down more than 1% the next day and up 4.11% five days out, so a trader with a tight exit would have been out before the count scored it as higher.
The decade mix changes. Bullish haramis on SPY ran 19, 60, 27 and 21 in the 1990s, 2000s, 2010s and 2020s. Almost half of them, 60 of 127, came in the 2000s, a decade in which SPY closed 49% and then 57% below an earlier peak, so the full-period figure leans on one stretch of history.
Related
The inside bar page covers the range-based version of this pause, which four in ten harami shapes on SPY also met. The bullish engulfing page describes the pattern that reverses the harami’s order, a big body covering a small one. And the candlestick patterns overview places the harami among the other two-candle and three-candle formations, including the morning star, whose middle candle plays a similar pausing role.
Read a harami as the market going quiet after a big day, not as a turn. If I use one at all, I want the next few sessions to prove it, and I size the position for the fact that the SPY leans in these counts were small enough to be chance.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.