Volume and What It Confirms

The short answer

Volume is a count of how much changed hands on one venue during one interval. That is the entire measurement.

It does not tell you whether buyers or sellers were in control, because every unit traded had one of each. It does not tell you who transacted or why. And because the same instrument trades in many places, the number on your chart is a local figure that cannot be compared with the figure on someone else's chart.

What a volume bar counts

For each interval, the venue adds up the quantity executed — shares, contracts, coins — and plots the total. Volume is therefore a measure of participation: how much activity occurred, not what that activity intended.

Three distinctions matter, and confusing them is the source of most volume mistakes.

Measure	What it counts	Where it lives
Volume	Quantity executed in the interval	Completed trades
Depth	Quantity resting at each price	The order book
Open interest	Contracts currently open	Derivatives positions

Depth is intent that has not executed and can be withdrawn. Open interest is exposure that has not been closed. Volume is the only one of the three that is a record of completed transactions — and it is silent about which side was the aggressor, unless your data feed separately reports trades at the bid and the ask.

The "buying volume versus selling volume" split you see in some tools is usually that aggressor classification, inferred from whether a trade printed nearer the bid or the ask. It is an inference, it is approximate, and it is not part of the volume figure itself.

Volume is a local number

A single instrument rarely trades in one place. In U.S. equities, activity is dispersed across many exchanges and off-exchange venues; a single exchange's volume is a share of the whole, and that share has changed substantially over the years as market structure changed. In foreign exchange there is no central exchange at all — trading runs across dealer platforms, electronic communication networks and single-bank streams, so any "FX volume" figure is a figure for whatever subset the source can see.

Three consequences follow directly.

You cannot compare across venues. A spike on one exchange may be a spike in that exchange's share rather than in total activity. The instrument may be no busier than usual.

You cannot naively sum venues. Depending on how trades are reported, the same economic transaction can appear in more than one dataset, and the composition of a consolidated figure differs from feed to feed. Summing tapes that were built on different conventions produces a number with no clean interpretation.

Thresholds do not transfer. A volume level that means "unusual" on one venue means nothing on another. Any rule involving a specific volume number is a rule about the venue whose feed you built it on.

This is also why relative volume — this interval against the same interval on comparable days, on the same feed — is more informative than any absolute figure. It holds the venue constant and asks the only answerable question: is participation here unusual for this feed?

What "confirmation" does and does not mean

The standard claim is that a move "confirmed by volume" is more trustworthy than the same move on low volume.

The defensible part is small and worth stating exactly: a move accompanied by unusually high participation means more of the market transacted at those prices than usual. That is a fact about the interval. It tells you the move was not a thin drift through an empty book, which is genuinely useful when you are deciding whether a level was tested by anyone.

What it does not establish:

• Direction. High volume on an up move means buyers and sellers were both extremely active. Someone sold every unit that was bought.
• Who. Volume is anonymous. Attributing it to "institutions" or "smart money" is a story added afterwards.
• Continuation. Heavy participation appears at the start of sustained moves and at the exhaustion of moves that end immediately. The bar looks the same in both cases.

Confirmation in this sense is descriptive, not evidential. It says two things happened together. The trap is the retrospective version: when a move continues, the volume is cited as having confirmed it; when it fails, the volume is re-read as a "blow-off". A concept that explains both outcomes after the fact is explaining neither in advance. If you use volume in a rule, write the threshold and the window down before you look, exactly as you would with any indicator.

Fragmented and decentralised markets

The limits get sharper in crypto and other venue-fragmented markets.

Reporting is not standardised. Centrally listed markets have reporting obligations behind their published figures. Many crypto venues publish self-reported numbers with no comparable obligation, and the definition of what counts — internal transfers, matched trades between related accounts, incentive programmes that reward volume — varies by venue.

Wash trading is a known concern. Activity that moves between related accounts inflates a venue's figure without representing genuine transfer of risk. You generally cannot detect this from a chart, which is the point: the volume bar looks identical.

On-chain and off-chain are different populations. A token's on-chain transfer volume includes movement that is not trading at all — bridging, custody moves, internal reshuffling — while the price you are trading may be set on a venue whose activity never touches the chain.

Perpetual and spot volumes are not interchangeable. They are different instruments with different participants, and derivative volume can dwarf the underlying spot market without implying anything about it.

The practical position: in these markets, treat volume from a single venue as a weak signal about that venue's own activity, and be sceptical of any figure aggregated across venues you cannot inspect. The crypto market structure lesson covers the wider version of this problem.

Where volume earns its place

Used modestly, it does three jobs well.

1. Sanity-checking a level. A test of support on very low participation is weak evidence that the level was examined at all. That is a statement about how much you should update, not about direction.
2. Flagging regime changes. A sustained shift in typical participation — a session that is persistently busier or quieter than its own recent norm — often accompanies a change in liquidity conditions, which affects spreads and slippage before it affects anything else.
3. Estimating execution cost. Thin intervals are where your own order moves price. This is the most reliable use of volume on the list, and the least discussed.

Notice that none of these is a directional signal. Volume tells you about the conditions you are trading in. Price tells you what happened. Keeping those two jobs separate is most of what it takes to use volume without over-reading it.

What to do with this

Check what your volume feed actually covers before you build anything on it — one venue, a consolidated tape, or an aggregate of sources you cannot audit. Then use relative volume on that feed rather than absolute numbers, compare like sessions with like, and state any threshold in advance.

And when you catch yourself saying a move was "confirmed by volume", finish the sentence honestly: more units traded than usual. Everything you were about to add after that was interpretation.