SQMATESQMATE
Cup and handleChart patternBacktest

We opened 30,000 cup-and-handles: the base wants to be long, the handle short

Only look at cups whose handle ended quickly — the longer it ran, the worse every measure got, and waiting never fixed it. The base is the opposite: cups that sat longer produced bigger moves. Ten years of bars, counted.


You have probably seen a shape on a chart that looks like a cup: price drifts down, rounds out a base, and after a while climbs back to the height it started from. Then, just under that height, it dips again briefly — the small handle on the cup.

Textbooks stop there. On a real chart you get stuck immediately. How long does a base have to be before it counts as long? How long is a long handle?

So we counted. From ten years of bars across Korean, US and crypto symbols we pulled the thirty thousand-odd cups SQMATE actually found, counted how many days each one spent at the base and how long its handle ran, and followed what price did over the next 60 and 120 days.

Four words and you can read to the end

The same drawing as in the pattern library (/patterns).
The rim is the height of the cup's two edges. From the left edge, around the base, to the right rim is the cup; the small dip after it is the handle. Going above the right rim is the breakout. The shaded band is where price sat near the bottom, and "the length of the base" in this article means the number of days in that band.

How we counted — only two decisions

① Which day we measured from. The day three days after the right rim forms. That is the day SQMATE first puts the cup-and-handle card on screen. Measuring only from there means we use nothing but what was knowable at the time — no quietly picking winners with hindsight.

② What we compared against. A number on its own says nothing. Is "up 9% in 60 days" good? So we compared each cup with other days in the same symbol and the same year that sat a similar distance below the high. That is the vs control column in the tables.

  • 0.50 means the same as those days.
  • Above 0.50 means better; below means worse.

By that ruler the cup and handle as a whole scores 0.48. Entering on the shape alone came out essentially the same as an ordinary day at a similar height. A slightly disappointing number — and the whole article starts there, because what separates cups inside that average is everything.

① A longer base means a bigger move — confirmed

The shaded band is the stretch spent at the base. The cup on the right held near the same price far longer, and grew deeper and wider for it.
Days at the baseCupsCup depthBest rise in 60 daysReached the targetReached a breakout
0–94,50520.9%8.9%15.7%67.0%
10–198,19321.9%8.3%14.3%61.2%
20–349,50124.0%8.8%15.4%58.5%
35 days or more9,02029.7%11.2%19.4%56.4%
"Days at the base" counts days when price sat in the bottom quarter of the cup. "Target" is the textbook objective — one cup depth above the rim.

31,219 cups · 6,349 symbols · real bars from 2016-01 to 2026-06 · computed 2026-08-20. A record of the past; it promises nothing ahead.

Cups that spent 35 days or more at the base rose the most within 60 days (11.2%); the other groups managed 8–9%. They also reached the target most often.

And the same table says why. Look at the "cup depth" column: as the base lengthens, the cup deepens with it — 20.9% to 29.7%. A bigger cup is built, so the move that follows is bigger. Divide the rise by the cup depth and the four groups come out almost identical.

A long base does not magically produce more upside. A price that stayed pressed for a long time builds a bigger, deeper cup, and what follows scales with the cup. The base sets the size of the move in advance.

The same table also carries the price of it. The share that reached a breakout fell as the base grew longer (67.0% → 56.4%). Choosing cups that can run big also means more that fail. You could not have both.

SOFI24.01.1724.12.03
rimfloorright rimdetectedbreakout
50 days at the basecup depth 31%handle 4 days (2% of the cup)60% after 60 daysbest 81% within 120 days
Fifty days at the base. The handle was a short four days, and price carried on after clearing the right rim. This is one chart that worked — far more cups of the same shape gave way instead.

② What to look for in the handle

The handle is the one stretch of a cup and handle where there is time to watch. The cup is already built; the breakout has not come. We asked whether those few days tell you anything about what follows.

On the left, a bar with a long tail below the body — price was pushed down and came back within the day. On the right, nearly the same price revisited over several days.
The handle on day 5CupsReached a breakoutGave way in the endUp after 120 days
Same price revisited9,07964.5%31.0%55.1%
Not revisited16,78752.3%43.6%50.1%
Has a lower wick16,33757.8%37.9%51.7%
No lower wick9,52954.5%41.2%52.2%
Both5,69765.8%29.8%54.9%
Neither6,14750.2%45.8%50.4%
Split purely on how the handle looked on day 5 — nothing that was not visible on screen that day. "Same price revisited" means the three lowest points sat within 1% of each other; "has a lower wick" means at least one bar whose tail below the body was more than half the bar's length.

31,219 cups · 6,349 symbols · real bars from 2016-01 to 2026-06 · computed 2026-08-20. A record of the past; it promises nothing ahead.

Handles that revisited the same price did much better. 64.5% reached a breakout, against 52.3% for the rest. Failures split 31.0% against 43.6%. Price being held at the same level again and again reads as sellers below that level actually thinning out.

A lower wick on its own made almost no difference — 57.8% against 54.5%. One tail does record that somebody absorbed supply that day, but it did not say as much as the same price being revisited over several days. With both present the figure is 65.8%; with neither, 50.2% — 15 points between the top and bottom of the table.

One more thing. Handles with revisited support produced smaller moves. Quiet, tight handles come from quiet symbols in the first place. Roughly shaken handles carry both the chance of running and the chance of breaking.

③ The longer the handle, the worse the odds

Handle length (vs cup)CupsHandle daysvs controlBest rise in 60 daysReached a breakout
0.00–0.1017,35470.54811.1%67.4%
0.10–0.207,808210.4488.6%53.6%
0.20–0.334,023370.3696.3%49.6%
over 0.332,034510.2944.8%39.4%
"Handle length (vs cup)" divides handle days by cup length. 0.1 means the handle is a tenth of the cup. The median across the sample is 0.087 — by the textbook, a handle is far shorter than its cup.

31,219 cups · 6,349 symbols · real bars from 2016-01 to 2026-06 · computed 2026-08-20. A record of the past; it promises nothing ahead.

Compare handles that finished inside a tenth of the cup with handles that ran past a third of it. Every column collapses in the same direction: vs control from 0.548 to 0.294, breakouts from 67.4% to 39.4%. The cups themselves (their depth) are almost identical across the four groups. The only difference is how long the handle took.

"So what about the handle I am holding right now?"

That table hides a trap. A handle's final length is only knowable once it is over. To someone on day 12 of a handle, "long handles did badly" is no help at all.

So we inverted it. We collected only the cups that reached a given day with neither a breakout nor a breakdown yet, and measured forward from that day — using nothing but what was knowable then.

Handle on dayCups not finished yetBroke out in the endGave way in the endvs control
day 330,75760.7%35.7%0.477
day 525,86656.6%39.2%0.500
day 1018,11951.7%42.2%0.515
day 1513,00149.7%41.8%0.527
day 209,59948.3%40.2%0.534
day 305,15643.6%35.0%0.540
day 402,98634.2%28.8%0.546
"Handle on day" counts days since the right rim formed. "Cups not finished yet" is how many had neither outcome by that day.

31,219 cups · 6,349 symbols · real bars from 2016-01 to 2026-06 · computed 2026-08-20. A record of the past; it promises nothing ahead.

Waiting does not improve the odds. Looking forward from day 3 of a handle, 60.7% eventually reached a breakout. From day 10, 51.7%. From day 20, 48.3%. From day 40, 34.2%. A handle that had already failed to go anywhere was closer to the group that never would.

This is why SQMATE caps handle length. Past 35 days the cup-and-handle card comes off the screen (for cups shorter than about 100 days it comes off at a third of the cup length instead). The shape has not vanished — the grounds for calling it a cup and handle have.

One thing to be clear about: the tables in this article deliberately keep the cups that fail that cap — you cannot see why a cap is needed unless the cups it catches are still in the table. Of the thirty thousand, 4,358 (14%) would not appear as a card on today's screen.

The worst of the three endings

How the handle endedCupsvs controlBest rise in 60 daysUp after 120 days
Cleared the rim18,6720.59212.9%64.2%
Lost the floor11,4410.2884.4%26.2%
Nothing by day 601,1060.1802.6%21.3%
Split only on how the handle ended. "Nothing by day 60" means sixty days that neither cleared the rim nor lost the floor.

31,219 cups · 6,349 symbols · real bars from 2016-01 to 2026-06 · computed 2026-08-20. A record of the past; it promises nothing ahead.

Breaking down is not the worst ending. Nothing happening is. Cups that lost the floor scored 0.289; cups where nothing happened for 60 days scored 0.180 — a best rise of 2.7% over 60 days, quieter than an ordinary day at a similar height. Money tied up while nothing happens.

DKNG25.02.0425.11.11
rimfloorright rimdetectedgave way
38 days at the basecup depth 34%handle 22 days (16% of the cup)-32% after 60 daysbest 2% within 120 days
A handle that dragged 22 days and then lost the floor. The line marked "floor" is the level that keeps price in the upper part of the cup; the cup's story ends on the day it closes beneath it.
DIS24.03.2125.04.10
rimfloorright rimdetectedday 60
45 days at the basecup depth 27%handle 60 days (36% of the cup)-6% after 60 daysbest 2% within 120 days
A cup that passed the same conditions and ended differently — its handle ran past 60 days. Compare the width of the handle band with the width of the cup. Four months went by without breaking and without going anywhere.

The shape names a place; the story supplies the force

That is the shape's half of the story, and the shape already showed its limit: the cup and handle as a whole scored 0.48. On shape alone it is nearly the same as the control.

So what separates the cups that ran from the ones that stayed pretty drawings lies outside the shape. It is the market's attention — whether the symbol has a fresh catalyst, whether the theme that catalyst belongs to is in favour right now, which phase its sector sits in. A textbook cup on a symbol nobody is watching has no buyers and stays a rounded picture. The same shape resolves differently on a symbol whose theme is running.

The shape tells you where to look. The force that pushes price out of that place comes from the catalyst and the theme. The odds are better when both are present — that is the conclusion of watching this pattern for years, and the tables above measured only the first half of it.

So the sentence this article leaves you with is one. Only look at cups whose handle ended quickly. Long base, short handle — measuring just those two filters out most cups. What is left is the part the catalyst and the theme answer.

What this looks like on the SQMATE screen

These rulers are not confined to the article; they are the ones the screen uses.

  • The card first appears three days after the right rim forms, and that date

stays on the card — it does not roll forward to today. Add the symbol to a watchlist and you can count what day of the handle it is on.

  • Past 35 days of handle the card comes down (for short cups, at a third of

the cup length). The tables above are the grounds for that cap.

  • The card disappears the moment price breaks out, because clearing the right

rim makes that the new high and leaves no "handle". A card vanishing is not a bad sign — it means the thing you were waiting for happened.

today under one chip. Shapes still forming are a separate filter.

  • Add a symbol to a watchlist or holdings and you get an email on the day a

good pattern completes on it. Build your own per-symbol conditions in the note screen and the browser notifies you the moment they are met.

symbols you found here back over that is the order this article suggests.

We opened 30,000 cup-and-handles: the base wants to be long, the handle short