2006issue C021-6
A five-by-five grid that accepts or rejects a cup
A rounded bottom, also called a cup, is a decline to a low, a quieter sideways stretch, then an advance back toward the prior high. This article restates that shape as a five-by-five log-price and time grid so recent bars either meet the occupancy rules or they do not, before volume or a breakout story is added.
- A rounded bottom, also called a cup, can be stated as occupancy rules on a five-by-five log-price and time grid.
- Logarithmic versus arithmetic scaling, bar versus line style, periodicity, and the chart width-to-height ratio can make the same series look more or less cup-like.
- Upper-side cells that must contain no closes keep the middle hollow, and specified lower cells must hold a bar low so the trough is actually visited.
- The scan is a shape filter for daily and weekly bars, not a buy signal. Volume and any handle stay outside the occupancy rules, and later bars can drop a name from the watchlist.
The cup as a spoken shape
A rounding bottom, also called a cup, is described as a decline to a low, a quieter sideways stretch, then an advance back toward the prior high. That quieter middle is base building: the stretch from the low after the first decline to the start of the right-side advance, treated as a constructed pause rather than a finished signal.
Sometimes a short, low-volume pullback appears after the right rim, before an upward break of the right-rim high. That optional pullback is a cup with handle. The identification rules treat the handle as extra context, not a required cell.
Shorter, steeper recoveries with a more volatile middle are sometimes called tower bottoms. They are still grouped with cups when that middle is not very volatile, even though the sides form faster than a textbook multi-week cup.
Why two charts of the same series can disagree
The same price series can look more or less cup-like depending on whether the y-axis is logarithmic or arithmetic, and also depending on bar versus line style, periodicity, and the chart width-to-height ratio.
A semilog chart uses arithmetic time and logarithmic price. The identification procedure works on natural logs, so a cup on this scale usually remains a cup on an arithmetic scale, but not always the reverse.
How the five-by-five grid is built
The identification procedure first replaces each bar's high, low, and close with its natural logarithm. It then takes the highest high of the last 20 bars as the right rim and walks backward to the first prior bar whose high is at or below that rim after a bar that exceeded it.
Between those two time anchors the lowest price becomes the floor. Five equal log-price bands and five equal time bands then form a 25-cell grid. Each cell height is one-fifth of the log-price span. That rectangle is the five-by-five grid used for the occupancy test.
Which cells must stay empty and which must be visited
Pattern recognition, in this construction, is a mechanical scan that tags whether recent bars occupy and leave empty the required cells of the grid over a fixed lookback.
A candidate is treated as cup-like only if designated upper-side cells contain no closes and if at least one pair of specified lower cells each contains a bar low. A closing-price blank is a grid cell that must contain no closes. Those empty upper cells keep the cup hollow instead of filled by mid-range trading, and the required lows make sure the trough is actually visited.
Volume is not part of the occupancy rules.
What the scan keeps, and what later bars can take away
The mechanical test is meant to scan large universes on daily and weekly bars and to keep hits on a watchlist, because later bars can drop a name. Many individual cups fail early in long declines. Cups that appear as a broad decline starts to recover are treated as names to study if the tape turns up.
All readings on this track · 14 readings
- 1990Constructing falsifiable reversal patterns from price structure
- 1995Cup-with-handle construction: confirm the cup, the handle, then the breakout
- 1995Cup-completion cheat before the handle breakout
- 1998Constructing rounded-bottom cups as testable entries
- 1999Rounded bottom landmarks, invalidation, and breakout rules
- 2003Constructing rounded bottoms, triangles, and pennants
- 2006A five-by-five grid that accepts or rejects a cup
- 2006NTRI and the 2005 bowl breakout case
- 2006Rounded-bottom screens, first-try breakouts, and Fibonacci retracements
- 2007Constructing rounded bottoms as multi-year bases
- 2011Early semi-cup construction from the left rim and base
- 2011Early rounded bottom recognition on a locked log-price grid
- 2013Rounded turns as slope-first trade hypotheses
- 2017Evaluating a rounded bottom as a testable payoff structure