Volume and On-Balance Volume

Every oscillator on a chart is computed from price. Trading volume is not. It measures how many participants acted, which is genuinely separate information, and it is the reason volume confirmation is such a persistent theme in technical work.

A three percent advance on triple the usual volume and a three percent advance on a third of it are the same price change produced by very different amounts of participation. One says many people agreed to transact at those prices; the other says very few did.

One caveat is worth stating early, because it is routinely got wrong: every share traded has both a buyer and a seller. There is no such thing as more buyers than sellers. What changes is which side is more urgent, and volume measures the total activity rather than the imbalance.

Absolute volume is not comparable between securities: a mega cap trades tens of millions of shares on a quiet day and a small cap trades fifty thousand on a busy one. Relative volume normalises against each security's own baseline, which is what makes it screenable across a market.

US equity volume is U-shaped. The first thirty minutes and the last thirty carry a disproportionate share of the day, with the closing auction the single largest event. The middle of the session is much quieter.

The construction assumes an up close means the day's volume was net buying, which is the simplification noted above. What it actually tracks is volume weighted by the sign of the close, which is a cruder thing than the name suggests.

Its level is arbitrary, since it depends on when the accumulation started. Only its direction and its divergence from price carry information, and comparing OBV values between two securities is meaningless.

The reading it supports is a comparison of slopes: price making higher highs while on-balance volume does not describes advances happening on progressively less participation, which is the same observation volume confirmation makes in a different form.

A raw volume figure is not comparable to anything. Ten million shares is enormous for one security and a quiet morning for another, so volume is almost always read relative to that security's own history.

Intraday, the comparison has to account for the shape of the session. Volume is heavily concentrated at the open and the close, so a figure at 10 a.m. compared against a full day's average will always look small. The correct comparison is against the volume normally accumulated by that time of day.

On-balance volume adds the whole day's volume to a running total when the close is up and subtracts the whole of it when the close is down. That single rule is both its appeal and its principal weakness.

The assumption is that a higher close means the day's volume was accumulation and a lower close means distribution. A day that rose a hundredth of a percent on heavy volume adds all of it; a day that fell by the same amount subtracts all of it. The measure is therefore extremely sensitive to closes near unchanged.

• The absolute level of OBV is meaningless, since it depends on when the calculation started.
• Only the direction and the divergences against price carry any claimed information.
• Variants such as the accumulation-distribution line weight by where the close fell within the day's range, which addresses the sensitivity but adds an assumption of its own.
• None of these can distinguish buying from selling, because every share traded has both.