Fake bullish candle: a green candle on a down day, and why crypto charts almost never print one

A fake bullish candle closes above its own open, so the chart draws it as a bullish candle, but it still closes below the previous candle's close, so measured close to close, the period still closed down. That's the whole definition: both conditions on the same candle. The term is a translation of 假阳线 ("fake yang line"), slang from mainland China's stock market, where traders ask some version of "the index fell today, so why is the candle red?" (red is the up colour there).

On crypto charts it barely exists. On 30 October 2026 we pulled 365 daily candles each for BTC, ETH, BNB and SOL from Binance's public market-data endpoint, plus 1,000 one-hour and 1,000 four-hour BTC candles, and counted every candle that met both conditions. The total was zero. The reason is the open: a Shanghai-listed stock gets a fresh opening price every morning and can open far from yesterday's close, while in our samples the open was never more than 0.0161% away from the previous close.

Two conditions on one candle

Written out so you can check a candle against it:

• Condition one: close above open. That's simply what makes a candle bullish. The colour is just paint your charting software applies on top (more in Bullish vs bearish candles).
• Condition two: close below the previous candle's close. Compare today's close with yesterday's and today is down.

Each one on its own is ordinary. Plenty of candles close above their open; plenty of days close lower. A fake bullish candle is the two landing on the same bar. One answer on the Chinese Q&A site Cofool (叩富问财), from 2013, describes how it typically happens, in our translation: if the stock opens well down, then closes above that open without getting back above the prior day's close, you get a red candle on a day the price fell. Which colour means "up" varies by market and by your settings; Green up or red up? covers that, so we'll leave it there.

Not every answer on that page is as careful. Another reply there boils it down to "the open being lower than the close makes it a yang candle", which is only condition one. By that reading every bullish candle would be fake. We count a candle only when both conditions hold.

Condition two hides a choice you make yourself: which candle is "the previous one". On an hourly chart it's the hour before, no argument. On a daily chart it's yesterday, but where yesterday starts and ends depends on your chart's time zone. Binance's API splits days at 00:00 UTC by default (08:00 in Beijing), and only splits at Beijing midnight if you set it to UTC+8; Crypto daily candle close time goes through the difference. Change the split and "yesterday's close" becomes a different number, so the same stretch of trading could qualify under one split and not the other.

Where it comes from: a stock market that reprices every morning

For the pattern to form, the open first has to land a fair way below yesterday's close, leaving room to climb without getting all the way back. The Shanghai Stock Exchange's trading rules (we read the consultation-draft text) set the day out like this: an opening call auction from 9:15 to 9:25, continuous trading from 9:30 to 11:30 and 13:00 to 14:57, and a closing call auction from 14:57 to 15:00. The opening price is the first trade of the day, and it comes out of that call auction.

Between 15:00 and 9:15 the next morning, nothing trades. Whatever news lands overnight gets priced in one go at the opening auction, so the open can sit a long way from yesterday's close. Open low enough, recover part of it during the day, fall short of the previous close, and you have one.

On that market it isn't rare. A September 2020 market report in Hangzhou's Dushi Kuaibao listed six dates that year when the Shanghai Composite printed one (3 February, 25 February, 13 March, 30 March, 12 June and 23 July), and the headline was about yet another, the previous Friday.

A year of Binance candles, counted one by one

The method is blunt. Binance's documentation points public market-data requests at data-api.binance.vision; we called /api/v3/klines, which per the endpoint docs returns each candle as an array with the open in position 2 and the close in position 5. Every candle was compared with the one before it. Close above open and below the previous close counted as a fake bullish candle; close below open and above the previous close counted as a fake bearish one. We also logged how often the open differed from the previous close at all, and the largest gap. Data pulled at 18:05 Beijing time on 30 October 2026.

• BTCUSDT daily, 365 candles (opens from 2025-10-31 to 2026-10-30, days split at UTC midnight): of 364 adjacent pairs, the open differed from the previous close in 172, by at most 0.0026%. Fake bullish candles: 0.
• ETHUSDT daily, 365 candles, same range: 189 of 364 differed, at most 0.0032%. Fake bullish: 0.
• BNBUSDT daily, 365 candles, same range: 178 of 364, at most 0.0087%. Fake bullish: 0.
• SOLUSDT daily, 365 candles, same range: 192 of 364, at most 0.0161%. Fake bullish: 0. Fake bearish: 0.
• BTCUSDT one-hour, 1,000 candles (2026-09-18 19:00 to 10-30 10:00 UTC): 460 of 999, at most 0.0016%. Fake bullish: 0.
• BTCUSDT four-hour, 1,000 candles (2026-05-16 20:00 to 10-30 08:00 UTC): 467 of 999, at most 0.0026%. Fake bullish: 0. Fake bearish: 0.
• BTCUSDT daily, 365 candles, with timeZone=8 (days split at Beijing midnight): 174 of 364, at most 0.0013%. Fake bullish: 0. Fake bearish: 0.

These numbers cover these pairs over these windows and nothing else. The last candle in each set was still open when we pulled it, and fake bearish candles were only counted in the last three sets. What we can say is that none turned up in this sample, not that one never will.

Here's what "differed" looked like, copied as returned: BTC's daily candle for 3 November 2025 opened at 110540.69 against a previous close of 110540.68; ETH on 1 November 2025 opened at 3848.0 against 3847.99; BNB on 6 November 2025 opened at 959.32 against 959.33. The first two opened 0.01 higher, the third 0.01 lower.

When the open sits on the last close, there's no room

Crypto spot trading doesn't stop between candles. One closes and the next opens straight away, with no overnight break and no morning auction to reprice. The counts show it: on the BTC daily chart, 192 of 364 pairs had an open exactly equal to the previous close.

Those can be ruled out on arithmetic alone. The close would have to be above the open and below the previous close, but the open is the previous close. They can't both be true.

For the rest, a fake bullish candle would have to close inside the sliver between the open and the previous close. Take the BNB candle: open 959.32, previous close 959.33, so the close would need to land above 959.32 and below 959.33, a window 0.01 wide. The widest window anywhere in our data was SOL's 0.0161%, under two hundredths of a percent. A full candle's close landing in a gap that narrow, approached from below, didn't happen once in a year of counting.

What we didn't test: the first candle after an exchange maintenance halt, the first candle of a new listing, futures candles, and any other pair. We have no data on how the open relates to the previous close in those cases, which is why we say "almost never" rather than "never".

The same point spills over into other patterns. Any rule that says "opens below the previous close" or needs a gap between two candles is working with a difference of a few decimal places on these charts, or none at all. The "opens below the prior close" part of a bullish engulfing was, on these dailies, a gap of at most about 0.016%, and roughly half the time zero. The gap a morning or evening star wants, and the gap-based rules in candlestick cheat sheets, come up rarely for the same reason. Their other conditions still work as written.

What the candle records, and what it doesn't promise

Treated as a specimen, it records three finished facts: the open was below the previous close; within the candle, price rose from the open for a stretch; and by the close it still hadn't reached the previous close. The second says someone was buying after the open. The third says that, against yesterday, the period still lost ground. Which of those matters more is a question the candle doesn't answer.

Commentary tends to go a step further. That 2020 Dushi Kuaibao piece read it as a sharp gap down on bad news, with price neither extending lower nor recovering the day's loss, and added that more back-and-forth "tends to follow". Everything before that last clause describes the day that already happened. The last clause looks forward, and the piece gives no sample for it: no count of candles, no count of how many were followed by a reversal. We'll take the description and leave the forecast.

Green candle, red percentage: three things to check

Since our samples didn't contain a single fake bullish candle, if an exchange chart shows a bullish candle next to a negative percentage, it's worth ruling out these before reaching for the term:

1. Where the percentage starts from. The change figure on a market page and the open of today's daily candle need not share a starting point, so they can disagree. The 24-hour change versus the daily candle is covered in Crypto daily candle close time.
2. Which time zone splits the day. A UTC-midnight split and a Beijing-midnight split give two different "previous closes". We counted under both and got zero fake bullish candles on BTC dailies either way, but the same day can come out up under one split and down under the other.
3. Which colour means up. Check your colour setting before reading a red candle the way a Shanghai trader would; see Green up or red up?

FAQ

Is a fake bullish candle a bullish or a bearish sign?

It only records what already happened that day, so we don't attach a direction to it; the reasoning is in the section above on what the candle records.

Does it count as a bullish candle or a bearish one?

By shape it's bullish: it closed above its open. Measured close to close, the day was down. The "fake" refers to that mismatch between shape and result; it doesn't mean the candle is secretly bearish. The only thing separating it from any other bullish candle is the second condition, a close below the previous close.

What's a fake bearish candle, and do crypto charts have them?

It's the mirror image: the close is below the open, so it's drawn as a bearish candle, yet the close is above the previous close, so the day was actually up. It also needs the open to sit well away from the previous close. We counted fake bearish candles in three of our samples (SOL daily, BTC four-hour, and BTC daily split at midnight Beijing time) and found zero in each.