Symmetry — how equal is equal?
The claim is that a double top whose two highs are close in price is a better example than one whose highs are far apart. As a description of how typical an example is, that is uncontroversial. As a rule you can apply, it immediately needs a number nobody supplies.
07.02.02 already ran into this and said so plainly: the highs are never level on a real chart, no published tolerance exists, and the judgement is yours and must be made in advance. The same applies here. ₹120 and ₹119 is less than 1% apart. Is ₹120 and ₹114 still a double top? Some readers say yes, some no, and the chart does not care which.
What makes this more than pedantry is 07.01.04's measurement: widening that tolerance from 2% to 5% roughly triples the number of double tops you find in the same data. The criterion is not a quality filter so much as a dial controlling how many patterns exist.
Duration — the criterion with a real idea inside it
The claim is that patterns taking longer to form are more significant. Two highs a fortnight apart is a smaller thing than two highs five months apart.
There is something genuine here and it is worth naming: a longer pattern involves more sessions, and therefore more participants, and therefore more people who dealt at that price. 04.06.01's argument scales — a price area tested across months has more sacks behind it than one tested across a week.
It is also, again, unquantified. Longer against what? The honest version is comparative rather than absolute: this pattern took eleven weeks, that one took nine days, and the first involves more of the market's history than the second. That is a defensible sentence. “Patterns under four weeks are unreliable” is not, unless somebody counted.
Volume — checkable, and unchecked
This is the interesting one, because unlike the other two it makes a claim that could be settled.
The textbook signature for a double top is that the second peak trades less than the first: the same price is reached with less participation, which is read as the buying that drove the first peak having thinned out. For a double bottom the mirror claim is that the second low trades less than the first, or that the breakout trades more.
That is a real, falsifiable statement about numbers that exist. You can look at any double top you have identified and check it in a minute. Almost nobody does, and the criterion survives as a thing people say rather than a thing people test.
It is also the only one of the three that brings information from outside the price line. Symmetry and duration are both read off the same closes that produced the shape. Volume is a separate record — which is exactly why 04.05.02's card kept it as its own filter.
- First peak
- ₹120, on 3,40,000 shares
- Second peak
- ₹119, on 1,90,000 shares
- Recent comparable sessions
- 1,80,000 to 2,60,000 shares
- What the first peak's volume was
- well above the comparable range
- What the second peak's volume was
- inside it, at the lower end
- What this establishes about the next session
- nothing
A criterion you can check and do not is worth less than one you cannot check at all, because you believed it anyway.
Why these three do not add up to a score
The obvious next step is to combine them. Symmetrical, long, with the right volume signature: three out of three, so a better pattern. Books do this, software does this, and it is the error 04.05.02 refused by name.
Four written answers make a reading inspectable; they do not vote a trade into existence. The same holds for three. Adding criteria that have never been measured individually does not produce a measurement, and a pattern scoring 3 of 3 has not been shown to behave differently from one scoring 1 of 3 — by anyone, as far as Granthara can establish.
What the three are genuinely for is description. They let you say precisely what kind of example you are looking at, and they make your reading inspectable by somebody who disagrees. That is the same job the four-filter card does, and it is a real job.