Mat Hold Pattern — 78% Continuation on a Sample That Barely Exists

In the candlestick pattern catalog, 78% continuation is a top-tier result — higher than bearish engulfing, higher than the evening star, higher than almost anything with its own name. Mat hold has that number. It would be cover-page material for the catalog, if not for one figure sitting right next to it: 52 occurrences across 4.7 million studied candles. Bulkowski, the author of the study, comments on his own statistics for this pattern with the word "Wow" and the honest admission that at this sample size, the results are likely to be wrong or to change substantially once more cases appear. This is an article about a pattern whose percentage is impressive and whose denominator invalidates it — and about what that pairing of numbers teaches about reading trading statistics.

What the Formation Looks Like

Mat hold is a five-candle uptrend-continuation pattern — a stricter, tougher cousin of rising three methods:

• Trend before the pattern: up.
• Candle 1: tall white. A strong demand impulse, a full-fledged trend candle.
• Candle 2: small black, closing HIGHER than candle 1's close. In the classic Japanese description, the second candle opens with a gap up — the main distinguishing feature versus rising three methods.
• Candle 3: small, any color.
• Candle 4: small black. Candles 2-4 form a gentle downward drift — a correction whose bodies stay above the low of the first candle.
• Candle 5: tall white, closing above the high of the four preceding candles. Demand returns and closes out the pattern with a breakout.

The psychology is textbook: after a strong impulse, supply tries a counterattack but can't even reach the base of the impulse across three sessions — the correction is shallow, lazy, built on small bodies. The fifth candle shows demand was only catching its breath. The difference from rising three methods is the higher perch of the correction (the second candle's close above the first candle's close), which makes the pattern look "stronger" visually — and rarer in practice, since there are more conditions to satisfy.

And here's this blog's usual disclaimer, doubled this time: five-candle structures with such precise conditions do happen on crypto spot markets, but no one has measured them there — and the version with a gap on the second candle's open doesn't occur at all on a 24/7 market. Everything below applies to US stocks on the daily interval.

What the Numbers Say

Statistics from Bulkowski's tests (~4.7 million daily candles, US stocks — not crypto, not intraday):

• Continuation: 78% of cases. Formally, one of the best directional results in the entire 103-pattern catalog.
• Sample: 52 occurrences. That sentence deserves a second read. The author needed a special tool just to find these cases in a multi-year database, and he warns outright that the numbers "will probably be wrong." At n=52, a 78% result has a confidence interval wide enough to drive a truck through — actual reliability could sit anywhere from below 65% to above 85%.
• Frequency: 93/103. One of the rarest patterns in the catalog. You could scroll through hundreds of charts and never see one.
• Performance rank: 86/103. And here's the second blow: even when the pattern does continue the trend, the average 10-day move afterward is among the weaker ones in the catalog. A high directional hit rate, but little meat on the breakout.
• Best average 10-day move: −7.21% — in a bull market, on a downward breakout (rank 2/103). A familiar irony from this series: the best measured move after a bullish continuation pattern occurred exactly when the pattern failed. Bulkowski himself suspects this figure wouldn't hold up with a larger sample.
• Price-target achievement: 67% at best (bull market, downward breakout).

There's also a tip from the encyclopedia that sounds like a joke but isn't: expect a price reversal once mat hold completes. The pattern closes with a tall white candle breaking the top of the structure — and statistically, that's exactly when a correction tended to arrive. Taller-than-median patterns carried further; expect the breakout to be upward.

How to Trade It (and How Not To)

How NOT to play it: build a strategy around a pattern with 52 data points. Not because 78% is a bad number — because it's a number without a foundation. If your plan is "I enter on a mat hold because it's 78% reliable," your plan is standing on a sample smaller than the number of trading sessions in a quarter.

Scenario 1 — mat hold as trend confirmation, not an entry trigger. What's valuable about this setup regardless of the statistics is the information: the pullback after the impulse was shallow and listless, supply had no real argument. If you're already in a position aligned with the trend, mat hold is an argument for holding it — not a reason to add leverage.

Scenario 2 — if you do enter, use the tidbits, not the legend. The preferred setup from the data: a taller-than-typical pattern (height matters), an upward breakout (the more common one), entry on a close of the fifth candle or on a breakout of its high. And immediate humility: since the data show a pullback right after the pattern completes, don't enter with full size right at the top of the fifth candle — it's more sensible to wait for a shallow pullback or enter with partial size.

Scenario 3 — respect a downside breakout. A close below the low of the pattern isn't "noise, because the pattern is strong" — it's a measured spot after which, in bull markets, an average move of −7.21% over 10 days followed. A mat hold that breaks down is a better signal than a mat hold that works.

Stop loss and target. The natural stop: below the low of the first candle (the base of the impulse — losing it invalidates the whole shallow-pullback narrative). Target: the nearest resistance, with partial profit-taking. A performance rank of 86/103 is clear: don't project long runs from a pattern that, on average, delivers a short one.

Myth vs Measurement

Myth: "Mat hold is one of the most reliable continuation patterns — 78%!" Measurement: 78% from a sample of 52, with the study's own author warning the numbers are probably wrong. A percentage without a denominator is marketing, not statistics.

Myth: "A strong-looking pattern means a strong move afterward." Measurement: performance rank 86/103 — the bottom quarter of the catalog. Mat hold is more often right about direction than it pays for being right. For comparison: rising three methods continues 74% of the time and also has a low performance rank (94/103) — this whole family of "shallow pullback" patterns looks beautiful and rides short.

Myth: "Rarity makes a pattern elite." Measurement: rarity makes it unmeasurable. 52 cases is too few for anything beyond a hypothesis — the same mechanism as the kicker (116 cases, a "strongest pattern" legend). Patterns whose numbers actually mean something occur hundreds or thousands of times.

Myth: "Once the pattern completes, the trend takes off immediately." Measurement: the encyclopedia says the opposite — expect a pullback right after mat hold completes. Whoever enters with full size on the close of the fifth candle is statistically buying a local exhaustion top.

Example Scenario

Stocks in a healthy uptrend. Monday: a tall white candle on volume. Tuesday-Thursday: three small candles, each a bit lower, but all their bodies staying well above Monday's low — and Tuesday even closes above Monday's close. Friday: a tall white candle breaks the week's high — mat hold complete. What does the data say? That continuation is likely (to the extent you can trust 52 cases), that the move will likely be modest, and that a pullback is probable right after the structure completes. The plan that matches the numbers: if you want a long, wait for a shallow pullback toward the correction's bodies, stop below the low of Monday's candle, target at the nearest resistance. If price instead closes below the low of the pattern — no debate, the structure just busted, and a busted mat hold in a bull market has historically been one of this statistic's best short signals.

Quick checklist:

• Is there an uptrend before the pattern, and is the first candle tall and white?
• Is the second candle small, black, closing above the first candle's close?
• Does the correction (candles 2-4) drift downward while bodies stay above the low of candle 1?
• Is the fifth candle tall, white, closing above the high of the structure?
• Do you remember the sample: 52 cases — trade small at most?
• Are you prepared for a pullback right after the pattern completes?
• Is the stop below the base of the impulse, target modest (performance rank 86/103)?

Mat hold and the kicker together make the catalog's best teaching duo: the kicker teaches that a dramatic picture doesn't guarantee good statistics, and mat hold teaches that impressive statistics don't guarantee a real sample. 78% on 52 cases and 51% on 5,000 cases are not the same kind of knowledge — the second number tells you more, even though it sounds worse. Before you trust a percentage, ask about the denominator. That question costs five seconds and saves real money.

FAQ

Rafał — Krypto Bez Ściemy

Trader and founder of Strefa Tradingu. He’s been breaking crypto down on YouTube for years — no hype, no signals, with a focus on market structure and risk management.

kryptobezsciemy.plInstagramYouTubeKlub Zero