OBV — On-Balance Volume
The answer
OBV keeps a running total. On a day when the close was higher than the previous close, it adds the whole day's volume. On a lower close, it subtracts the whole day's volume. It measures whether volume has been arriving more on up days or on down days. It is a cumulative line, so its absolute level means nothing — only its direction and its disagreements with price carry information.
Why this costs you money
OBV is taught with a sentence that sounds profound and is not true: "volume precedes price." From that sentence people build the belief that a rising OBV means large institutions are quietly accumulating, and that price will follow.
Here is what OBV actually does. It takes the day's total volume — every share traded, by anyone, for any reason — and assigns 100% of it to the buyers if the close was 1 paisa higher, or 100% of it to the sellers if the close was 1 paisa lower.
Every trade has a buyer and a seller. There is no such thing as a day's volume belonging to one side. OBV's core assumption is a convention, not a measurement.
The cost comes in 2 forms.
A false accumulation story. A stock drifts sideways. OBV rises. A trader concludes that somebody large is accumulating and buys. In fact the stock closed up 12 times and down 9 times over the period, with roughly similar volumes, and the rising line is the arithmetic of 12 minus 9. There is no institution. There is a counting rule.
A single day that rewrites the line. OBV adds the whole day's volume. On a results day or an index rebalancing day, a stock can trade 15 times its normal volume. That 1 day can move OBV further than the previous 3 months combined, and if that day happened to close up by a small amount, OBV now shows a large "accumulation" that was actually a one-off liquidity event.
OBV is not a window into institutional behaviour. It is a running total of volume signed by the direction of the close, and the signing rule throws away most of the information in the day.
How it works
The rule is the simplest in this cluster.
- If today's close is higher than yesterday's close:
OBV = previous OBV + today's volume - If today's close is lower than yesterday's close:
OBV = previous OBV − today's volume - If today's close is the same: OBV is unchanged
That is all of it. There is no averaging, no smoothing and no scaling.
Three consequences follow directly and they are the whole indicator.
1. The absolute value is meaningless. OBV's starting point is arbitrary — usually 0 or the first day's volume, depending on how far back your chart goes. Two charting platforms will show different OBV values for the same stock. Only the slope matters.
2. Size of the price move is ignored. A day that closed up 0.1% and a day that closed up 9% contribute identically, as long as their volumes were the same. A trader looking at a rising OBV line cannot tell whether the up days were meaningful.
3. One day can dominate. Because the whole volume is added, a single extraordinary volume day carries more weight than dozens of ordinary ones.
The 3 things people look for
Confirmation. Price makes a new high and OBV makes a new high. This says the advance is happening on volume that is arriving on up days. It is the least controversial use and it adds the least.
Divergence. Price makes a new high and OBV does not. This says the second advance had relatively more volume on its down days. It is the most interesting use and it is where all the honest caveats live.
Breakout confirmation. Price breaks a level and OBV breaks its own recent high on the same day. This is the most practical use, because the question "did volume show up for this breakout?" is a real question with a real answer.
What it tells you, and what it does not
OBV tells you whether the volume of the last N days arrived more on up-closes or down-closes.
Five limits, and they are unusually severe for a widely used indicator.
It cannot identify who traded. There is no buyer/seller split in exchange volume data. The signing rule is an assumption.
It ignores the size of moves. A tiny up-close and a huge up-close are identical inputs.
It ignores intraday activity entirely. A stock that rose 6% during the day and gave it all back to close down 0.1% contributes its entire volume to the sell side. Everything that happened inside the session is discarded.
It is dominated by outliers. Index inclusion days, results days, block deals and expiry days produce volumes many times normal. Those days set the shape of the OBV line.
It cannot be compared across stocks or across periods. The level is arbitrary and the scale is the stock's own volume.
The clean statement: OBV asks a fair question — is volume arriving on up days or down days? — and answers it with a method that discards the size of the move, the path within the day and the identity of the traders.
That does not make it useless. It makes it a rough measure that requires a rough interpretation. The mistake is treating a rough measure as a precise one.
The decision rule
Use OBV for 1 job: checking whether volume supported a price event you already identified from price.
If price broke a level today and OBV also broke its own recent high, volume arrived on the up day. That is weak confirmation and it is worth having.
If price broke a level and OBV did not move much, volume did not arrive. Reduce your confidence in the breakout.
Treat OBV divergence as a place to look, never as a signal. A divergence requires price confirmation — a broken trend structure — before it means anything.
Unless the OBV move was produced by 1 or 2 extraordinary volume days. Check the volume bars. If a single day did most of the work, the line is reporting an event, not a trend in participation.
Try this now
This takes 5 minutes and it shows you the noise problem directly, which is the thing you most need to know about OBV.
- Open 2 daily charts set to 1 year. Pick a very liquid large company from your own watchlist, and a smaller company you also follow that trades far less volume. If you trade in India, use an NIFTY 50 name and a small-cap name.
- Add OBV to both, and make sure the volume bars are visible on the price panel.
- On each chart, find the 3 largest volume days of the year. On the OBV line, look at what those 3 days did to the line's shape.
- Now measure the damage. On the smaller company, estimate how much of the year's total OBV movement came from those 3 days. On the large company, estimate the same.
- Finally, on both charts, find 1 clear OBV divergence — price making a higher high while OBV made a lower high. Check what price did over the following 20 sessions.
What you should see. On the small company, the OBV line will often have 2 or 3 visible steps in it, and those steps are single days. A large share of the entire year's OBV movement can come from 3 sessions. On the liquid large company the line will be much smoother, because no single day is a large fraction of the year's volume.
That is the honest verdict on OBV in thin markets, and you have just measured it rather than been told it. OBV on an illiquid stock is a record of a handful of large orders.
In step 5, on most charts, the divergence will be followed by a mixture of outcomes. Do this on 10 divergences and you will find that a large share of them led nowhere. That is not a reason to discard divergence. It is a reason never to trade one without price confirming it first.
Three real cases
1. Joseph Granville, Granville's New Key to Stock Market Profits, 1963 (United States) — the tool and the record of the man who popularised it Joseph Granville popularised On-Balance Volume in his 1963 book and built a large following as a newsletter writer through the 1970s and early 1980s. Granville is also remembered for a specific and public failure: he turned strongly bearish in the early 1980s and remained so through the beginning of one of the longest bull markets in American history. The lesson is not that OBV is worthless. It is that the person most associated with volume-leads-price was, at the most important turning point of his career, wrong in the direction his own indicator was supposed to protect against.
2. The NIFTY 50 and NSE small-caps, 2020 (India) — the same indicator, 2 different noise levels Through 2020 the Indian market fell sharply and then recovered strongly. On the NIFTY 50 index and its most liquid constituents, daily volume is large and reasonably stable, so OBV moved smoothly. On smaller NSE-listed companies, where daily volume can be a small fraction of a large-cap's and can vary by a factor of 10 from one day to the next, OBV lines show large single-day steps. The same indicator, on the same exchange, in the same year, produced a usable line on one set of stocks and a step function on the other. Liquidity, not the indicator, determined which.
3. Index inclusion and rebalancing days (India and United States) — one day that rewrites the line When a stock is added to a major index, funds that track that index must buy it, and they typically do so around a specific date. Volume on that day can be many multiples of normal. OBV adds that entire volume to whichever side the close happened to fall on. A single mechanical, non-discretionary flow can therefore create what looks like a massive accumulation signal. Nothing about anybody's opinion changed. A rule required a purchase.
The question that resolves it
A novice sees OBV rising while price is flat and asks: is somebody accumulating?
An expert asks: how many days built this line?
If the rise came from 40 sessions of slightly positive closes on ordinary volume, it is a genuine, if weak, signal about participation. If it came from 2 sessions of enormous volume that happened to close up, it is a record of 2 events, and calling it accumulation is a story attached to arithmetic.
You can answer this question in 10 seconds by looking at the volume bars. Almost nobody does.
What would make this wrong
The claim is that OBV measures the direction of volume arrival using a crude signing rule, that it is dominated by outlier days, and that it is much noisier on thinly traded instruments.
You would falsify the noise claim with the exercise above. Run it on 10 liquid and 10 illiquid charts. If the illiquid OBV lines are as smooth as the liquid ones, this article is wrong about your market.
You would falsify the divergence claim by collecting 30 OBV divergences and checking outcomes 20 sessions later. If most of them resolved in the divergence's direction without needing price confirmation, believe your count over this page.
The honest limits are 4.
First, "volume precedes price" is not entirely a myth. There is a real body of work on volume and returns, and some studies find volume carries information . What is unsupported is the leap from that to OBV specifically, with its particular signing rule.
Second, OBV's crudeness is partly addressed by alternatives. Accumulation/ Distribution and Chaikin Money Flow weight the volume by where the close fell within the day's range, which uses information OBV discards. They have their own problems. If OBV's signing rule bothers you, those are the direction to look.
Third, the counting exercise measures how often, not how much.
Fourth, nothing here shows that any volume indicator produces profit. It shows what OBV measures and where its construction stops being trustworthy.
In India
This is the market where OBV's weaknesses matter most, and the reason is liquidity.
Indian volume is concentrated. A large share of NSE cash market turnover happens in a relatively small number of names. Below the most liquid few hundred companies, daily volume falls sharply. In those names, a single institutional order or a single block deal can be several times the normal daily volume.
OBV adds that whole order to one side. So an OBV line on an Indian mid-cap or small-cap is frequently a record of a handful of large trades rather than a measure of broad participation. The line will look like a staircase. Each step is one decision by one desk.
Delivery volume is published in India and it is more useful than total volume. NSE and BSE publish the deliverable quantity — the number of shares that actually moved between demat accounts rather than being squared off intraday. Total volume includes intraday trading that leaves no position behind. Delivery percentage is available on the exchange websites and on most Indian broker platforms.
This is a genuine Indian advantage and OBV does not use it. A version of OBV built on delivery volume would answer the accumulation question far better than one built on total volume, because delivery volume is closer to what "somebody bought and kept it" actually means. Most Indian platforms do not offer this. You can approximate it by checking the delivery percentage on the days that produced your OBV move — if a large OBV step happened on a day with low delivery percentage, it was intraday activity, not accumulation.
Expiry days distort volume. Indian derivatives expiries produce large cash market volume as positions are settled and rolled. Those days are mechanical. OBV signs them by the close like any other day.
In the United States
Deeper markets, and 2 different distortions.
Volume is fragmented across venues. US equities trade on many exchanges and alternative venues, and a meaningful share of volume executes off-exchange . The consolidated tape captures most of it, but what your charting platform shows depends on its data source. Two US charts of the same stock can show different volume, which means they show different OBV.
Extended-hours volume is usually excluded. US shares trade before 9:30 and after 16:00 Eastern time. Most platforms exclude that from the daily volume bar. On a results day, a large amount of trading happens in extended hours and does not enter OBV at all, while the price change it caused does enter — as a close-to- close move that gets signed and applied to the regular-hours volume.
That combination is genuinely misleading. OBV records the direction of a move that happened in a session whose volume it did not count.
Volume is deep and stable in large caps. For an S&P 500 constituent, no single ordinary day is a large fraction of a year's volume, so OBV lines are smooth and the outlier problem is confined to results days and index events. This is why OBV looks better in American textbooks than it does on an Indian mid-cap chart. The textbooks were written about the deepest equity market in the world.
Where they differ, and what that tells you
The difference is signal-to-noise, and it is large.
An OBV line on a liquid US large-cap is built from thousands of participants every day, so a single order barely registers. An OBV line on an Indian mid-cap is built from far fewer participants, so a single order can dominate a month.
What that tells you is where to trust OBV and where to stop. OBV divergence on a heavily traded index or large-cap is a real, if weak, observation about participation. The same divergence on a thinly traded stock is very likely to be an artefact of 2 or 3 orders, and treating it as institutional accumulation is a story about a number that a single desk created.
There is a second, more useful consequence, and it runs in India's favour. India publishes delivery volume and the United States does not have a directly comparable public figure. So the Indian trader who wants to know whether volume represented real accumulation has a data source the American trader does not. OBV throws it away. You do not have to.
The practical instruction differs by market. In the United States, use OBV as published and be careful around results days. In India, ignore OBV on thin names, and on any name check the delivery percentage on the days that produced the OBV move. The extra 20 seconds converts a crude indicator into a much better one, and it uses data that only Indian traders have.
Carry this
- OBV signs the whole day's volume by the direction of the close. That is a convention, not a measurement of buyers and sellers.
- Only the slope matters. The level is arbitrary and differs between platforms.
- Before believing any OBV move, look at the volume bars and ask how many days built it. In India, also check the delivery percentage on those days.