Hikkake pattern: trading the false breakout of an inside bar

The hikkake is a trap pattern. It starts with an inside bar, a session whose high and low both fall within the previous session's range. Inside bars attract breakout traders: when price pushes beyond one side, they jump in expecting a move. The hikkake is what happens when that breakout fails and price swings back through the other side of the inside bar, leaving the breakout traders stuck.

Unlike most candlestick patterns, the hikkake is modern. It was described by the technician Daniel Chesler in the early 2000s, and its name is a Japanese word for tricking or ensnaring. It ignores candle colors and bodies entirely and works on highs, lows and closes, so it applies equally to bar charts.

It also behaves differently inside Finabase. TA-Lib reports the hikkake in two stages, with two different values, and it fires far more often than any classic pattern. This page explains both stages, how the scanner and Signals tab treat them, and what the data says.

The setup and the trigger

A bullish hikkake unfolds over up to six bars:

1. A reference bar with a normal or wide range.
2. An inside bar: a lower high and a higher low than the reference bar. The market is contracting.
3. A false break lower: the next bar has both a lower high and a lower low than the inside bar. Breakout sellers act. TA-Lib marks this bar +100.
4. The confirmation: within the next three bars, a close above the inside bar's high. The breakout has failed and reversed through the far side. TA-Lib marks that bar +200.

The bearish setup is the mirror image: after an inside bar, a bar with a higher high and a higher low prints −100, and a close below the inside bar's low within three bars prints −200.

The exact rules Finabase uses

Finabase runs TA-Lib's CDLHIKKAKE on daily bars:

Details that shape the output:

• No size, color or trend tests. Only highs, lows and one close matter. There are no averages, which is why TA-Lib needs just five bars of history for this function.
• Two separate markers. The setup and its confirmation land on different bars. A confirmed hikkake therefore shows up twice: +100 and then +200 a few bars later.
• A new hikkake overwrites a pending confirmation. If a fresh setup forms on the same bar that would have confirmed the old one, only the new setup is reported. The old confirmation is lost.
• Most setups never confirm. In our data 31% of bullish setups and 23% of bearish setups went on to print a confirmation.

Why the trap works

An inside bar is a moment of indecision: neither buyers nor sellers could extend the previous session's range. Many traders place orders just beyond both sides, ready to join whichever way price breaks. When the third bar breaks lower, those sell orders fill, and short-term traders are positioned for a decline.

If the decline fails to develop and price turns back up through the inside bar's high, everyone who sold the breakdown is wrong, with stops typically above the inside bar. Their buying to cover adds fuel to the reversal. The idea behind the pattern is that a failed breakout can say more than a successful one, because it reveals where the crowd was positioned and forces it to exit.

How the scanner and gauge treat ±100 and ±200

Finabase does not distinguish the two stages anywhere except on the chart.

• Scanner. The candlestick scanner sorts every detection by sign. A +100 setup and a +200 confirmation both appear on the bullish list; −100 and −200 both appear on the bearish list. Check the chart to see which stage a listed stock is in.
• Signals tab. In the Candlestick gauge on a symbol page (for example AAPL's signals), each of the 61 patterns votes from its most recent non-zero value within the last eight bars: positive is Buy, negative is Sell. The size of the number is ignored, so a +200 counts exactly like a +100. And because the latest value wins, a bullish confirmation followed by a fresh bearish setup a few bars later flips the vote to Sell.
• How often it votes. In our sample, a hikkake value sat in the eight-bar window on 58.5% of all stock-days. Most other patterns are Neutral on nearly every day, so the hikkake is one of the few inputs that is almost always taking a side. Since the gauge computes (buy − sell) / (buy + sell + min(neutral, max(buy, sell))), a frequent voter can tip a close reading.

Two real hikkakes

IBM, July 2026 (confirmed). IBM had fallen about 32% in ten sessions. The July 20 bar ranged from $208.82 to $216.28, and July 21 was an inside bar from $209.18 to $214.97. On July 22 IBM broke lower, to a high of $212.34 and a low of $204.73: a +100 setup. The confirmation needed a close above $214.97. July 23 closed at $206.65 and July 24 at $214.19, just short. On July 27, the third bar, IBM closed at $216.28 and TA-Lib printed +200. From the setup close the stock was 14.7% higher ten sessions later; from the confirmation close it gained about 9.1% over the following seven sessions.

Tesla, March 2025 (unconfirmed). Tesla had dropped about 26% in ten sessions. March 4 ranged from $261.84 to $284.35, March 5 was an inside bar from $267.71 to $279.55, and March 6 broke lower ($260.02 to $272.65): a +100 setup. A confirmation required a close above $279.55 by March 11. The closes were $262.67, $222.15 and $230.58; on March 10 alone the stock fell 15.4%. Ten sessions after the setup, TSLA was 10.3% lower. A trader who treated the +100 as a buy signal took the loss; one who waited for +200 never entered.

What the data says

We ran CDLHIKKAKE over daily bars from October 30, 2024 to October 29, 2026 for 60 large US stocks and ETFs. It produced 1,268 bullish setups, 398 bullish confirmations, 1,450 bearish setups and 336 bearish confirmations, between 36 and 78 signals per symbol.

Setups alone carried no information. Both +100 and −100 behaved exactly like random days, which makes sense: at that point the breakout has not failed yet.

Confirmations showed a modest edge. The +200 bars rose 3.7 points more often than the baseline, and the −200 bars fell 6.3 points more often, with a negative average return. The bearish margin is a little over two standard errors from the baseline; the bullish one is within the range that chance alone could produce. Both are small in trading terms.

The study's own confirmation test agrees. Taking all bullish values together and requiring the next close to clear the signal bar's high, 482 cases rose over the next five sessions 60.4% of the time, measured from that confirmation close. The same filter on the bearish side (412 cases) gave 47.8%, no better than the baseline.

Hikkake versus similar setups

An educational outline

A common way to structure a hikkake trade, for study only:

1. Wait for the confirmation (±200). The data above suggests the setup alone is noise.
2. Enter on or after the confirming close.
3. Place the stop beyond the false breakout's extreme, the low of the +100 bar for a bullish hikkake. If price goes back there, the trap has failed.
4. Keep the time limit in mind. If no confirmation arrives within three bars, TA-Lib abandons the setup and so do most practitioners.

Frequently asked questions

What does a +200 hikkake value mean?

It is the confirmation bar: price closed beyond the opposite side of the inside bar within three bars of a +100 setup. TA-Lib adds 100 to the setup's value to mark it.

Is the hikkake bullish or bearish?

Either. A false break lower sets up a bullish hikkake; a false break higher sets up a bearish one. The sign of TA-Lib's output tells you which.

Why does Finabase show so many hikkakes?

Because the conditions are simple: any inside bar followed by a directional bar qualifies. On large US stocks that happens about once every nine sessions.

Does Finabase treat a confirmed hikkake as stronger?

No. The scanner and the Signals-tab gauge only use the sign, so +100 and +200 are treated the same. Only the chart lets you see which stage you are in.

Summary

The hikkake turns a failed inside-bar breakout into a signal: the break prints ±100 and a close back through the far side within three bars prints ±200. Finabase's scanner and gauge only read the sign, so they treat setups and confirmations alike, and because the pattern is so common it influences the Candlestick gauge on most days. In our data the setups were indistinguishable from random days, while the confirmations showed a small edge. If you use it, wait for the ±200 and know where the trap would fail.