The evening star is the bearish twin of the morning star: a strong up day, a small candle that gaps higher, then a hard down day that cuts deep into the first candle. The rally was going fine, it stalled, and then sellers took over. That’s the story, anyway.
Our scanner has logged 2,181 evening stars across 559 US stocks and ETFs over the past three years. Its bullish twin failed our test badly, so I was curious whether this one would too.
How it’s built
- Day one: a long up candle.
- Day two: a small body that gaps above day one’s body.
- Day three: a long down candle that closes well into day one’s body (at least 30% of the way, in TA-Lib’s definition).
If day two is a doji, it’s an evening doji star.
What happened next
For a bearish pattern you want the “higher” numbers to come in below the baseline.
| After a… | Cases | Higher 1 day later | Higher 5 days later | Higher 20 days later |
|---|---|---|---|---|
| Evening star | 2,181 | 55.0% | 53.4% | 53.8% |
| Evening doji star | 758 | 53.1% | 50.0% | 52.0% |
| Any day | 413,576 | 51.9% | 53.8% | 55.6% |
The day after an evening star, the stock was up 55% of the time, more often than usual. That’s the same snap-back I see after most big one-day drops. Over five days it’s basically the baseline. Over 20 days it’s about two points weaker than a random day, which leans the right way but is within the margin of error of about two points.
The doji version looks better: higher five days later only 50% of the time, almost four points below the baseline. But with 758 cases the margin of error is around 3.6 points, so I wouldn’t lean on it much.
After a rally, where it’s supposed to work
| Before the evening star | Cases | Higher 5 days later | Any day in the same spot |
|---|---|---|---|
| Rose more than 5% in 10 days | 756 | 56.2% | 53.7% |
| Moved less than 5% | 1,292 | 51.0% | 53.1% |
| Fell more than 5% | 99 | 57.6% | 54.0% |
This is the strange part. After a strong rally, the textbook setup, stocks kept rising more often than usual after an evening star. It did best, as a bearish signal, when the stock had been going sideways.
Confirmation
When the next day closed lower again, the stock was higher five days later 51.6% of the time, against 54.0% for any down day. That’s 2.4 points in the right direction, but on 964 cases the margin of error is about 3 points. Suggestive at best.
How I’d use it
Mostly as a reason to manage a position I already have, not to open a short. If a stock I own prints an evening star at a level it has struggled with before, I might sell a covered call or tighten my stop. Betting on a drop on the strength of this pattern alone isn’t something the data supports.
Recent evening stars
The scanner found 56 evening stars in October 2026, including SOLV and WRB on October 23 and CAT and CMI on October 22. See today’s on the candlestick scanner, and every pattern in Do Candlestick Patterns Work?
Data: daily prices for 559 US stocks and ETFs, October 2023 to October 2026, patterns detected by TA-Lib. Educational only, not financial advice.