Candlesticks · Three-candle patterns
Tasuki gap (upside and downside): a gap that holds on a pullback
How the upside and downside tasuki gap work, the equal-body rule TA-Lib adds, how it differs from gap three methods, and what 195 real cases did next.
On this page
A tasuki gap is a continuation pattern built around a gap, which Japanese traders call a window. In an upside tasuki gap, a white candle gaps up above the previous candle and the next session is a black candle that pulls back into that gap but does not close it. The downside version is the mirror image: a black candle gaps down, and a white candle rallies into the gap without closing it.
The idea is that an open window acts as support (or resistance on the way down). If the first pullback after the gap stalls inside it, the gap has done its job and the trend is expected to continue. Tasuki is a sash used to tie back kimono sleeves; Nison uses the term for this gap pattern and notes that it is rare.
TA-Lib detects both directions with one function and adds a requirement many textbooks leave out: the white and black bodies must be about the same size. This page covers both versions on one page, explains exactly where the tasuki gap ends and Upside and downside gap three methods: when a gap gets filled begins, and shows how 195 tasuki gaps on large US stocks behaved.
Anatomy of the tasuki gap
The upside version in three steps:
- Before the window. Any candle in an advance. Its real body sets the lower edge of the gap.
- The gap candle. A white candle whose whole body opens and closes above the prior body.
- The tasuki candle. A black candle that opens inside the white body and closes below the white candle's open, into the gap, but still above the top of the first candle's body.
The downside version swaps every direction and color: a black candle gaps below the prior body, and a white candle opens inside it and closes above its open, inside the gap, but below the bottom of the first candle's body.
The exact rules Finabase uses
Finabase runs TA-Lib's CDLTASUKIGAP on daily bars. "Near" is 20% of the average high–low
range of the five candles before the gap candle.
| Test | Upside (+100) | Downside (−100) |
|---|---|---|
| Gap | Gap candle's body entirely above the first body | Gap candle's body entirely below the first body |
| Gap candle | White | Black |
| Third candle color | Black | White |
| Third candle's open | Inside the white body | Inside the black body |
| Third candle's close | Below the white open, above the top of the first body | Above the black open, below the bottom of the first body |
| Body sizes | Gap candle and third candle bodies differ by less than "near" | Same |
Some consequences of the code:
The first candle's color does not matter. Only its body position is used to define the gap.
Shadows can overlap. As in all of TA-Lib's gap tests, the window is between real bodies. The third candle's low may dip below the first candle's body as long as its close does not.
The score follows the gap candle's color. TA-Lib returns the gap candle's color times 100, so every upside tasuki gap is +100 and every downside one is −100.
No trend test. Nothing checks that the gap continues an existing trend. In our data, 37.8% of upside tasuki gaps formed after a ten-session decline, which is not the setting the textbook describes.
Tasuki gap versus gap three methods: the exact difference
These two patterns start identically, a candle and a gap candle, and differ in how far the third candle travels:
| Feature | Tasuki gap | Upside and downside gap three methods: when a gap gets filled |
|---|---|---|
| First candle's color | Not checked | Must match the gap candle (white for upside) |
| Third candle closes | Inside the gap; gap stays partly open | Inside the first body; gap closed |
| Body-size rule | Gap and third bodies nearly equal | None |
| Third candle's open | Inside the gap candle's body | Inside the gap candle's body |
| Score | Gap direction: up = +100, down = −100 | First candle's color: white = +100, black = −100 |
Because one closes inside the window and the other inside the first body, a single bar can never satisfy both. Both are classed as continuation patterns; the tasuki gap is the shallow pullback that leaves the window partly open, and the gap three methods is the deeper one that closes it.
Why the open window matters
A gap marks a price range where no trading took place. Buyers on the gap day were willing to pay above the entire previous body without any negotiation in between. When the next session pulls back, the question is how deep the profit-taking goes.
If sellers manage only to push into the gap and the close stays above the earlier body, the buyers who drove the gap are still in profit and the gap remains a visible area of support. That is the tasuki logic: a pullback that respects the window. If the pullback closed the window completely, the gap-day buyers would be back to even, and the bullish argument would rest on something else.
The equal-body rule in TA-Lib adds a sense of balance: the pullback was about as large as the advance that preceded it, a strong counter-move that still failed to close the gap.
Confirmation
- Upside: the next session closes above the black candle's high, or the gap candle's high.
- Downside: the next session closes below the white candle's low.
- The gap is respected on later retests.
- The trend is real. A tasuki gap in the direction of a clear trend, confirmed by a moving average or the Relative Strength Index (RSI): how to read and use it staying above 50 (upside) or below 50 (downside), is the textbook setting.
Two real upside tasuki gaps
AMD, May 7, 2026. AMD closed at $355.26 on May 5, then opened at $409.49 on May 6, a window of about $54, and closed at $421.39 on a white body. On May 7 it opened at $417.07, inside the white body, and closed at $408.46: below the white open, deep in the gap but nowhere near the top of the May 5 body. The black body ($8.61) was close enough in size to the white body ($11.90) to pass TA-Lib's test. The next session closed at $455.19, and ten sessions later the stock was 10.1% above the pattern close. A strong trend, a huge window and a shallow pullback: the textbook case.
Meta, July 13, 2026. META surged on July 9 (a $47.49 white body to $631.48), gapped up to open at $660.34 on July 10 and closed at $669.21. On July 13 it opened at $661.38 and closed at $656.73, $3.61 into the $28.86 window. Every rule was met, and two sessions later the stock closed at a new high of $681.31. Then it reversed, fell through the whole window to close at $627.17 on July 22, and ten sessions after the pattern it was 9.6% lower. A gap that holds for one day has not necessarily become support.
What the data says
Over two years of daily bars (October 30, 2024 to October 29, 2026) for 60 large US
stocks and ETFs, CDLTASUKIGAP fired 195 times on 56 symbols: 103 upside and 92
downside.
| Group | Cases | Moved the pattern's way after 5 sessions | Average 5-session return | Average 10-session return |
|---|---|---|---|---|
| Any day, share higher (upside baseline) | 29,760 | 53.1% | +0.35% | +0.66% |
| Upside tasuki gap | 98 | 53.1% | +0.26% | +0.88% |
| Upside, confirmed | 19 | 52.6% | −0.07% | — |
| Any day, share lower (downside baseline) | 29,760 | 46.7% | +0.35% | +0.66% |
| Downside tasuki gap | 92 | 44.6% | +0.43% | +0.75% |
| Downside, confirmed | 11 | 36.4% | +0.78% | — |
- Upside: identical to the baseline. 53.1% higher after five sessions, the same share as all days, with average returns close to the baseline.
- Downside: slightly worse than random. Only 44.6% were lower a week later, and the average stock rose. In a two-year sample where large stocks mostly went up, bearish continuation patterns had the wind against them.
- Confirmation added nothing measurable, and the confirmed samples are small.
Other gap and continuation patterns
| Pattern | Third candle | Gap at the end |
|---|---|---|
| Tasuki gap | Opposite color, closes inside the gap | Partly open |
| Upside and downside gap three methods: when a gap gets filled | Opposite color, closes inside the first body | Closed |
| Side-by-side white lines: the twin-candle gap continuation | Second white candle with the same open as the first | Open |
| Upside gap two crows: the bearish pattern that leaves its gap open | Black, engulfs a small black above the gap | Open |
| Rising / falling three methods | Five candles: a long candle, small pullback, resumption | n/a |
An educational outline
How practitioners often use the pattern. It isn't a recommendation to trade.
- Insist on a trend. The tasuki gap is a continuation pattern; it needs something to continue. Skip cases that gap against the prior move.
- Mark the window. The top of the first candle's body (upside) is the level that must hold. A close beyond it turns the setup into a closed gap.
- Wait for a resumption close beyond the tasuki candle's extreme.
- Size from the far edge of the window. In wide-gap cases like AMD's, that distance can be very large, which shrinks the position for a fixed risk.
How Finabase shows the tasuki gap
- Chart markers. The Tasuki Gap study marks the third candle of each upside and downside tasuki gap on daily charts.
- Scanner. Upside gaps score +100 and appear on the bullish list; downside gaps score −100 and appear on the bearish list. The latest candle on a bullish-list stock is a black one, which is expected.
- Signals tab. The Candlestick gauge on a symbol page such as AAPL's signals gives each of 61 patterns one vote from its most recent reading in the last eight bars: an upside tasuki gap is a Buy vote, a downside one a Sell vote. The score (buy − sell) / (buy + sell + min(neutral, max(buy, sell))) reads Strong Sell below −0.6, Sell below −0.2, Neutral below 0.2, Buy below 0.6 and Strong Buy above.
Frequently asked questions
Is the tasuki gap bullish or bearish?
Both exist. The upside tasuki gap is bullish (+100) and the downside tasuki gap is bearish (−100). Each predicts that the existing trend continues.
What happens if the third candle closes the gap?
Then it is no longer a tasuki gap. If it closes inside the first candle's body and the first two candles share a color, TA-Lib calls it an upside or downside gap three methods.
Do the two bodies really need to be the same size?
In TA-Lib, yes, within 20% of the recent average daily range. Many textbook descriptions mention it as a typical feature rather than a requirement.
Does a gap between shadows count?
No. Finabase, through TA-Lib, measures gaps between real bodies only, so a gap on the chart's bodies may still have overlapping wicks.
Summary
The tasuki gap tests whether a fresh window survives its first pullback. The third candle pushes into the gap and stops, leaving the window partly open. TA-Lib adds an equal-body rule and scores the pattern by the gap's direction. Close it a little further, into the first body, and it becomes the gap three methods. On 195 cases in two years of large-cap data, neither direction beat an ordinary day. The pattern is a useful way to mark a level, and the level only matters once the trend resumes from it.
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 — CDLTASUKIGAP 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 CDLTASUKIGAP