What Is On Balance Volume (OBV)?
OBV, or On Balance Volume, is a running total of volume: added on bars that close higher and subtracted on bars that close lower. Its absolute value depends on where the count started, so only the direction and shape of the line mean anything.
OBV is the simplest volume indicator there is — Joe Granville published it in 1963 — and still the one worth learning first, provided you know which half of it to ignore.
How it works
One rule, applied bar by bar:
close higher than the last → add this bar's volume
close lower than the last → subtract this bar's volume
close equal → leave it alone
That is the whole indicator. There is no smoothing, no threshold and no setting.
The number on the axis is meaningless
Both lines above are the same indicator on the same data. One started counting at zero, the other at 500.
A running total has to start somewhere, and the starting point is arbitrary — it is wherever your chart’s data begins. Load two years of history instead of one and every value changes.
So the level tells you nothing and the shape tells you everything. Rising, flat, or falling, and whether that agrees with price. Anyone quoting an OBV value as though it were a measurement is quoting the length of their data feed.
The all-or-nothing problem
This is the real weakness and it follows straight from the rule.
A bar that closes one tick higher contributes its entire volume as buying. A bar that closes one tick lower contributes all of it as selling. There is no partial credit and no notion of how much higher.
On this chart the smallest up bar closed 0.0005 higher and had its full volume counted.
So two bars with identical volume — one that surged and one that barely moved — score the same, and a long grind of tiny alternating closes can move the line a long way for no reason.
That is what the Money Flow Index and Chaikin Money Flow exist to fix, each in a different way.
It is cumulative delta
The order flow page builds cumulative delta by adding a bar’s volume when it closes up and subtracting it when it closes down. That is this indicator’s definition, word for word.
So candle-direction delta and OBV are the same construction, arrived at from two directions thirty years apart. If you have one on the chart, the other is not a second opinion.
Real order flow — the kind built from bid and offer on every trade — is genuinely different. This is not that, and neither is the delta on that page.
The reading people run it for
Price went from 100.92 to 101.35 — a higher high. OBV went from 3.9 to 1.4.
The story: the second push was made on less volume than the first, so fewer participants were behind it.
The honest status: it is a description of two lines disagreeing, and it resolves either way. It is worth having as a question — why did that happen? — rather than as an answer.
Why the crudeness partly survives
An obvious question this page raises: if OBV throws away so much, why has it outlasted most of what replaced it?
Because there is nothing in it to tune. No length, no threshold, no smoothing. Every other page here has a settings section that ends with the same warning about choosing the parameter that fits the history you tested, and this indicator does not offer you the rope.
Because the failure modes are visible. When OBV misleads it is usually obvious why — a run of tiny up closes, or a gap the rule could not see. Compare that with an oscillator reading 78, which is wrong in a way you cannot inspect.
And because the question it asks is the right one, even though the way it answers is crude. “Did volume go the same way price did” is worth asking; the disagreement is worth investigating; and a blunt instrument that asks a good question beats a precise one that asks a bad one.
None of which makes it accurate. It makes it legible, which is a different virtue and, on a chart you have to read quickly, not a smaller one.
A worked example
Check volume is real for the instrument. In spot forex it is not, which makes this a tick-count indicator rather than a volume one.
Ask one question: is the line going the same way price is?
If yes, there is nothing to do. Agreement is the ordinary state and carries no signal.
If no, treat it as a reason to look harder, not as a reason to trade. Find the level, find the invalidation, and let the price chart give you the number.
The original data
Across our study of 24,971 trading videos, 54 cover OBV. The median one gets 8,475 views, 87% never pass 50,000, and the median length is 8.0 minutes.
The corpus carries description text for 53 of those 54 — nearly all of them — and across those 53, one mentions invalidation, failure, or what a bad read looks like.
When it fails
Sideways the line wanders
In a trading range the up and down closes roughly alternate, so the running total drifts on whichever side happened to carry more volume.
That drift looks like a trend in the panel and is an artefact of the counting.
The rule ignores size
Covered above and it is the structural one. One tick and one big move score identically, so a quiet market with a slight upward bias can produce an impressively rising line.
A gap breaks it
The rule compares close to close, so a bar that gapped down and closed above the previous close counts as buying — the gap itself is invisible to the arithmetic. That is the same blind spot the Chaikin Money Flow page ends on, arriving by a different route.
You found the divergence afterwards
Divergences that preceded a turn are obvious later. At the time they look exactly like the ones that preceded another leg up, and there are more of those.
Related
Volume is the raw number this accumulates, and what it does and does not say.
Order flow builds the identical line under a different name, and explains what the real version would need.
And Money Flow Index is the version that weights by how far price moved rather than counting every bar in full.
This is the crudest volume tool there is and I think that is close to a virtue, because you can hold the whole thing in your head. What I use it for is one question: while price was going up, was the line going up too? If it was not, I want to know why before I do anything. What I ignore completely is the number on the side of the panel, which is an accident of when the chart started.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.