Evaluating volume-weighted breadth extremes

A two-step classroom test

Editorial: TradersWeek treats volume-weighted market breadth as a two-step classroom test. First decide whether volume is concentrating in advancing or declining names. Then decide whether that concentration is still building or has already stretched far enough that a contrary reading, a component-smoothed variant, or a simpler up-volume share is the more falsifiable next check.

Market breadth is the mix of advancing versus declining issues, read here with the volume that accompanies each side rather than as a raw issue count alone. That comparison is volume-price analysis: activity on advances is set against activity on declines to test whether volume confirms or contradicts the direction of price. On-balance volume is the conceptual cousin of the same question, because it assigns volume to the prevailing price direction and asks whether buying or selling volume is carrying the tape.

Write the ratio, then keep the sign

A volume-weighted breadth ratio can be written as advances over declines divided by up volume over down volume. That construction is the same as average volume per declining issue over average volume per advancing issue.

Under the standard definition, a high reading means declining issues are more active than advancing issues and is treated as bearish. A low reading means the opposite and is treated as bullish. Some chartists invert the ratio so high values are bullish. The established published definition keeps high values bearish.

Editorial: keep the published sign convention fixed before you judge an extreme, or the second step of the test changes meaning.

When an extreme can flip

At excessive extremes, the same reading can be flipped and treated as a contrary signal because the volume imbalance is judged unlikely to extend much further.

Editorial: the flip is a second decision, not a rewrite of the first. The first decision is still whether volume is concentrating in advancing or declining names. The second is whether that concentration has already stretched.

Smoothing already on the record

Recorded smoothing choices include a 21-day average for intermediate work, a 55-day average for longer horizons, a 40-day average, and a 33 percent exponential update treated as roughly a five-week simple average. These are separate ways to slow the raw volume-weighted breadth ratio. They are not additional proofs that the first reading was correct.

Open-component smoothing

An open variant averages each of the four component series first, commonly over 10 days, then applies the ratio. That open-component smoothing is meant so one volatile session cannot dominate every term at once. The same construction can be computed on more than one exchange listing and the three readings can be combined.

Editorial: use this variant when the classroom question has shifted from the latest tape print to whether the four inputs still agree after a short average.

The 33 percent exponential on the record

A 33 percent exponential of the standard ratio was observed as mostly unhelpful for the 1985 advance, mostly neutral through 1985, somewhat bullish in late 1986, bullish into 1987 until 10 weeks before the crash, then bearish five weeks before that crash. It was still described as insufficient on its own.

Editorial: treat that sequence as a historical walk-through of one smoother, not as a claim that the exponential answers both classroom questions.

Up-volume share as a later check

Up volume as a percentage of total volume, smoothed with a three-week average, is treated as a short-horizon demand measure because the tape is said to need buying volume merely to hold even. That up-volume share was described as erratic in 1984 and 1986, more often bullish than bearish in 1985, mostly bullish in the first half of 1987, then bearish nine weeks before the 1987 crash and at its most bearish in more than four years three weeks before the crash.

Editorial: switch to this simpler gauge when the remaining question is short-horizon demand, not when you only want another restatement of the volume-weighted breadth ratio.

Choosing the next falsifiable check

Editorial: after the raw volume-weighted breadth ratio has answered where volume is concentrating, choose the more falsifiable next check rather than stacking every smoother as confirmation. If the reading looks stretched, a contrary treatment is one candidate. If a single session may be dominating the four inputs, open-component smoothing is another. If the question has narrowed to short-horizon demand, up-volume share is the simpler gauge. The archive workflow records those choices. It does not rank them as a finished trading rule.