Candlesticks · Three-candle patterns
Evening doji star pattern: a doji at the top of a rally
Learn to recognize the evening doji star, TA-Lib’s doji and gap tests, why Finabase’s own signals skip it, and what 50 real occurrences did afterwards.
On this page
Picture a stock that has been rising for weeks. One session it posts a big green candle. The next morning it opens higher again, yet by the close it has finished exactly where it started: a doji. On the third day it falls hard, erasing a good part of that big green candle. That three-day sequence is the evening doji star.
It belongs to the star family. The plain Evening star candlestick: how to spot a three-day top allows any small body in the middle; this version insists on a doji, a session whose open and close are practically the same. Candlestick tradition treats a doji after a long rally as a warning even before the third candle arrives, because it shows an eager market suddenly unable to make headway.
This guide covers the formation, the thresholds TA-Lib uses, a Finabase configuration detail that keeps the pattern out of the app's own signals, two real cases, and a count of how all 50 detections in our two-year study behaved.
Anatomy of an evening doji star
Three sessions, three roles:
The thrust. A long green body extends the advance. It needs to be larger than the stock's typical recent body, so that it reads as a strong day for that particular stock.
The doji. The following session's body is a sliver, and the whole sliver sits above the thrust candle's body. The session may have long shadows, showing that price swung in both directions, or almost none. Either way the close lands back at the open.
The rejection. A long red body that ends well inside the thrust candle's body. This is the candle that turns hesitation into a possible reversal.
A useful way to keep the family straight: if the middle candle is a small body of any kind, the formation is an Evening star candlestick: how to spot a three-day top. If it is a doji, it is this pattern. If the doji is cut off from both neighbors by price gaps that even the wicks don't close, it is a bearish Abandoned baby candlestick: the rare double-gap reversal.
TA-Lib's thresholds
Finabase uses TA-Lib's CDLEVENINGDOJISTAR function, which returns −100 on the third
session when every condition below is met. Averages follow TA-Lib's default candle
settings and cover the 10 sessions before the candle being checked.
| Session | Condition |
|---|---|
| Thrust | Green, with a body bigger than the 10-session average body |
| Doji | Body at most 10% of the 10-session average high-low range |
| Doji | Bottom of the doji's body higher than the top of the thrust body (wicks may overlap) |
| Rejection | Red, with a body bigger than the 10-session average body |
| Rejection | Close below the thrust candle's close by more than penetration × the thrust body; default 0.3 |
Where the code parts company with the books:
- No uptrend test. The source file itself says the pattern matters in an uptrend but leaves the check to the user. In our sample 58% of detections came after a 10-session rise.
- A shallow rejection is enough. The default asks for a 30% retracement of the thrust body. Nison's examples typically close past the halfway mark.
- Only one gap. The doji has to sit above the thrust; nothing requires a gap down into the rejection candle.
Why Finabase's tools stay silent
The rejection rule has a parameter, optInPenetration, measured in multiples of the thrust
body. Finabase's configuration passes 10, a figure inherited from the legacy app and kept
for exact compatibility. Ten thrust bodies below the thrust's close is a drop no
three-session pattern delivers, and none appeared in the 29,760 bars we checked. So
Finabase will not draw, scan for, or score this pattern. We re-ran the same function
at 0.3 for the examples below; the chart link at the top of this page does the same, and
the Morning star candlestick pattern: rules, examples and test results page shows the arithmetic in full.
What the three sessions say
The thrust day is the crowd at its most confident. Buyers pay up all session and the close is strong.
The doji is where that confidence meets resistance. The gap up at the open says demand was still there overnight, but at those prices sellers matched every buyer, and the session ended level. Many candlestick writers read a doji after a long advance as a sign of a tired market. After a long run, a market that cannot go up on a strong open is showing that the easy buying is done.
The rejection day settles it, at least for now. Price falls through the doji and deep into the thrust day's body, and everyone who bought on those two days is now at a loss. In a mature rally, that can start a wave of profit-taking.
It can also do nothing. Uptrends pause constantly, and a rally that has three sessions of indigestion often resumes. That is why most traders want more than the pattern itself.
Confirmation
The usual checks before acting on an evening doji star:
- A close below the rejection candle's low soon afterwards, the rule used in our test.
- A gap down the following morning that holds.
- Volume higher on the rejection session than on the thrust session.
- Context: the doji forms at a previous high, a resistance band or an obvious round number.
- Momentum diverging, for example an Relative Strength Index (RSI): how to read and use it that fails to confirm the new price high.
Be aware of a twist in our data: the confirmed cases did worse than the unconfirmed ones, though there were only 18 of them.
Two evening doji stars from the record
Morgan Stanley, March 26, 2025. The stock had gained 7.2% in ten sessions. On March 24 it rose from $122.31 to $124.27. On March 25 it opened at $124.90 and closed at $125.20, a doji above that green body. On March 26 it opened at $125.77 and fell to $122.20. The next day it closed at $119.00, below the pattern's low, which confirmed it. Five sessions on, it was 2.4% lower; ten sessions on, 8.6% lower, during the broad market decline of early April 2025. The same session produced an evening star on GE Aerospace, discussed on the Evening star candlestick: how to spot a three-day top page. Both worked, and both owed much of their success to a market-wide fall that started days later.
Cisco, May 19, 2026. Cisco had rallied 22.4% in ten sessions. It rose from $114.61 to $118.21 on May 15, printed a doji on May 18 ($118.80 open, $118.88 close), and dropped to $115.38 on May 19. The pattern even confirmed on May 20 with a close of $114.35. Then the advance resumed: five sessions after the pattern Cisco was 3.7% higher, and after ten it was up 9.6%, trading as high as $129.42. A powerful rally can digest a three-day stumble, confirmation and all. A stop above the doji's high would have capped the loss.
What the data says
Our study ran TA-Lib's evening doji star detector at penetration 0.3 over two years of daily bars (October 30, 2024 to October 29, 2026) for 60 large US stocks and ETFs, 29,760 day-samples in total. For a bearish pattern, the hit rate is the share of cases with a lower close five sessions later.
| Group | Cases | Lower after 5 sessions | Average 5-session return | Average 10-session return |
|---|---|---|---|---|
| Any trading day (baseline) | 29,760 | 46.7% | — | — |
| Every evening doji star | 50 | 44.0% | +0.93% | +1.48% |
| Confirmed, measured from the confirmation close | 18 | 33.3% | +1.78% | — |
The evening doji star was followed by a lower close slightly less often than an average day, and on average prices rose afterwards, by 0.93% over five sessions and 1.48% over ten. The confirmed subset was weaker still. Set against the 154 plain evening stars, which hit 47.4%, the doji requirement did not sharpen the signal in this sample.
How it compares with its neighbors
| Pattern | Middle session | Distinguishing feature |
|---|---|---|
| Evening star candlestick: how to spot a three-day top | Any small body | The looser parent pattern; 154 cases in our sample |
| Evening doji star | Doji | Body at most 10% of the average range |
| Abandoned baby candlestick: the rare double-gap reversal (bearish) | Doji | Gaps on both sides that the shadows don't cross either |
| Three black crows: meaning, TA-Lib rules and real examples | Red candle | Three falling red candles; no star at all |
| Morning doji star: the doji version of the bullish star pattern | Doji | The same pattern upside down, after a decline |
An educational outline for trading it
The steps below describe a common approach. They are for learning, not a recommendation, and the figures are hypothetical.
- Find a rally that is extended, ideally into resistance. TA-Lib won't check this.
- Look at the doji. Long upper shadows show buyers were rejected at the high, which many traders find more telling than a flat doji.
- Decide the trigger: the rejection close, or a later close under the rejection low.
- Mark the failure point just above the doji's high.
- Size so that a move to that point is an acceptable loss.
How Finabase shows it
- Charts. The pattern list includes Evening Doji Star. With Finabase's penetration of 10 it never places a marker; the chart link above passes 0.3 instead, so you can see the cases TA-Lib's default would flag.
- Scanner. The candlestick scanner files every detection by the sign of TA-Lib's output, so this pattern (always −100) would sit on the Bearish list. At Finabase's setting the list has nothing for it.
- Signals tab. On a symbol's Signals tab, such as AAPL's, the Candlestick gauge turns each pattern's most recent detection within eight bars into a Buy or Sell vote. With no detections, this pattern always votes Neutral.
Frequently asked questions
What is the difference between an evening star and an evening doji star?
Only the middle candle. An evening star accepts any small body there; the evening doji star requires a doji. Every evening doji star in our data also qualified as an evening star.
Does the doji need long shadows?
No. TA-Lib tests only the doji's body, against 10% of the average range. Long upper shadows are a bonus many traders look for, because they show a failed push higher.
How often did evening doji stars lead to a decline?
In our study, 44.0% of 50 cases closed lower five sessions later, compared with 46.7% of all days. The average five-session move was a gain of 0.93%.
Can I see this pattern in Finabase?
Not with the app's default configuration, which uses a penetration of 10 inherited from the legacy app. The chart link on this page sets it to TA-Lib's default of 0.3, which does show the pattern.
Summary
The evening doji star adds a doji to the evening star: a strong up day, a stalled day perched above it, and a strong down day. The doji is supposed to make the warning louder. On two years of large US stocks it did not. The 50 cases were lower a week later no more often than ordinary days, and the stock usually rose instead. Finabase's inherited penetration setting means its own tools never flag the pattern. Treat it as a question about an extended rally, not an answer, and know where you would be proven wrong.
Sources and further reading
- Steve Nison, Japanese Candlestick Charting Techniques (New York Institute of Finance, 1991)
- Thomas N. Bulkowski, Encyclopedia of Candlestick Charts (Wiley, 2008)
- TA-Lib source code — CDLEVENINGDOJISTAR and the default candle settings
- Finabase chart data — daily bars for 60 US stocks and ETFs, 2024-10-30 to 2026-10-29; patterns detected with TA-Lib CDLEVENINGDOJISTAR (penetration 0.3)