Breadth · Volume-weighted
Advance/Decline Volume
Every other breadth measure gives each issue one vote. This one weights them by the stock that changed hands: which matters only on the sessions where the two counts disagree, and those are the sessions worth finding.
A census, and a weighted vote
The advance/decline count treats every issue alike: a company worth a few million and one worth a few hundred billion each contribute one vote. That is the property that makes breadth data worth having next to a capitalisation-weighted index, and it is also a limitation. It cannot tell a move made in size from one made in a trickle.
Up and down volume fill that gap. The exchange reports the total shares traded in advancing issues and the total traded in declining ones, so the same session can be described twice: how many issues went each way, and how much stock went with them.
That is the honest case for the measure, and it is narrower than the literature suggests. Most sessions, the two series say the same thing; the disagreements are where the information is, and they are worth watching precisely because they are uncommon.
The convention that assigns each share a side
The whole of an issue’s daily volume is credited to one direction, decided by where it closed relative to the previous close. Nothing about the path within the session enters the calculation.
So a stock that fell for six hours, recovered in the last twenty minutes and closed a cent up contributes every share it traded to up volume. It is the same convention on-balance volume uses on a single instrument, and it is the reason both measures are best read as rough classifications rather than as measurements. A tick-based version — assigning each trade to the direction of the price change that produced it — is a genuine measurement, and it requires trade-level data that daily breadth statistics do not contain.
The ratio, and the thresholds attached to it
| Configuration | What it describes | What it cannot support |
|---|---|---|
| Both counts strongly one way | An unambiguous session: most issues moved, and most of the stock moved with them. | A forecast. This is the most common decisive configuration and it resolves in both directions. |
| Issues up, volume down | A broad advance in small size, with heavy trade concentrated in a few falling issues. | A conclusion about the market: check whether one or two very large names produced the volume split. |
| Issues down, volume up | A narrow advance carrying the stock, often a few large issues on results or an index event. | A breadth claim of any kind. This is the configuration a weighted index reports as a good day and a census reports as a poor one. |
| Volume ratio beyond nine to one | A "ninety per cent day", the published basis for several breadth-thrust claims. | The threshold as printed. Nine to one came from one exchange in one era; recompute it as a percentile of your series. |
The third row is the configuration worth learning to recognise, because it is the one that makes a weighted index and a breadth count tell opposite stories about the same day, and it is the situation the whole breadth section exists to make visible.
Why volume concentration makes this measure fragile
Issue counts and volume are distributed quite differently. On any exchange the number of issues is spread across thousands of names, while volume is concentrated: a small number of very heavily traded securities account for a large share of the total on any given session.
The consequence is that the volume split can be set by a handful of names, and on a day when one enormous issue trades on results, the up/down volume figure is substantially a statement about that company. The issue count cannot be captured this way; it takes thousands of issues to move it. So the volume version is more informative in principle and less robust in practice, which is the trade-off to hold in mind whenever the two disagree.
The practical habit follows directly: when the two measures diverge, look for the large issue before reaching for an interpretation. A divergence caused by one company’s results day is a fact about that company, and the temptation to read it as a statement about participation is exactly what makes this measure easy to misuse.
TRIN is this comparison in one number
Dividing the advance/decline ratio by the up/down volume ratio gives the Arms Index, which is precisely the relationship the figure above draws as two panels. A reading above one means volume was concentrated in the declining side relative to what the issue count would suggest; below one, the reverse.
That is a compact way to watch for the disagreements, and it comes with the standard warning attached to every ratio of ratios: the number is volatile, its distribution is skewed, and its familiar thresholds are era-specific. Watching the two series separately is less elegant and harder to misread, which is why this page draws them apart rather than together.
Frequently asked questions
What are up volume and down volume?
The total shares traded in issues that closed higher, and the total traded in issues that closed lower. They are published alongside the advance and decline counts, and they answer a different question: not how many issues took part, but how much stock moved on each side. That makes this the only breadth measure that is not a one-issue-one-vote census.
What does the volume version add?
It distinguishes a broad move made in small size from a narrow one made in large size. A session where slightly more issues advanced but the great majority of the volume was in the decliners is a genuinely different event from one where both agree, and the issue count alone cannot see the difference. Those disagreements are uncommon, and they are the only reason to keep this measure alongside the count.
How is a share assigned to a direction?
By where its issue closed relative to the previous close, the whole session’s volume in that issue goes to one side. So a stock that fell all day and closed a cent above the previous close contributes every share it traded to up volume. It is a convention, not a measurement, and it is the same convention that on-balance volume and its relatives depend on. Tick-based versions exist and require trade-level data.
What is the ratio people quote?
Up volume divided by down volume, sometimes on a nine-to-one basis: a session where up volume is more than nine times down volume is described as a "ninety per cent up day", and the mirror case as a ninety per cent down day. Clusters of those days are the basis of several published breadth-thrust claims. The observation is reasonable; the specific threshold is a number drawn from one exchange in one era, and it should be recomputed as a percentile of your own series.
Does it belong with TRIN?
It is one half of TRIN. The Arms Index divides the advance/decline ratio by the up/down volume ratio, so it is precisely a comparison of the two measures on this page, a reading above one means the volume was concentrated in the declining side relative to the issue count. If you find the volume version informative, TRIN is the compact way to watch the same relationship.
Which issues distort it?
The same ones that distort every breadth series, and one more. Closed-end funds and preferred shares move together on rate news; multiple share classes double-count a company. The additional problem here is that volume is far more concentrated than issue counts are: a handful of very heavily traded names can determine the volume split on their own, so this measure is much more sensitive to a few large issues than the census version is.
Is a cumulative up/down volume line useful?
Its shape, yes; its level, no, exactly as with the advance/decline line. A running total of net up volume depends entirely on when the accumulation began, so the number carries no information. What can be read is its direction against a price index, and divergences between the two. Anyone quoting the level of such a line is quoting an arbitrary starting date.
Where does the data come from?
The exchanges publish up and down volume with the daily breadth statistics, and data providers redistribute them. Historical series are generally licensed. Note that consolidated volume includes off-exchange trading, so a figure from one venue is a fraction of the total, the short-volume study on this site works through what a consolidated volume figure does and does not contain.