Volume and OBV: what they actually measure and what they do not

Volume has a reputation as the honest indicator, the one that cannot lie because it is real transactions. The first part is true; the second, less so. And the OBV, which is the most common way of reading it, works in a way almost nobody explains and that completely changes how it should be interpreted.

What volume is and why it is slippery in crypto

Volume is how many units of an asset changed hands in a period. On a traditional exchange it is reasonably clean data, because one centralised market publishes it.

In crypto no such single market exists. Every exchange publishes its own, and two problems follow. The first is that reported volume does not always correspond to real trades: wash trading — buying from yourself to inflate the figure — is a known and documented phenomenon on small platforms competing for ranking positions. The second is that adding up every exchange's volume does not give a better figure, it gives a worse one: it mixes clean data with inflated data.

The OBV and the thing almost nobody says

The OBV (On Balance Volume) is a running total: it starts at an arbitrary number and, for each candle, adds the entire volume if the close was higher than the previous one, and subtracts it entirely if it was lower.

Read that again, because everything is in there. It does not add a share proportional to the size of the move. It adds or subtracts the whole volume. A candle closing one hundredth higher contributes exactly as much as one that rises ten per cent.

The diagram above shows the consequence: two consecutive candles with nearly identical volumes and tiny closes in opposite directions make the OBV jump up and then down by the full amount. The OBV does not ask how much price rose, only whether it rose.

Two rules follow that save trouble. First: in sideways markets the OBV lurches meaninglessly, because closes alternate by cents and the whole volume switches sides every candle. Second: its absolute value means nothing. It depends on when the accumulation started. An OBV of 40 million is neither better nor worse than one of 2 million.

So what is it for?

One thing, and it does it well: spotting when price and volume stop telling the same story.

If price makes a higher high but the OBV does not follow, that advance happened with less participation than the previous one. It does not guarantee a fall, but it does say fewer people are pushing, and moves sustained by few people are fragile.

The same in reverse and, in practice, more reliably: price making a lower low while the OBV holds usually indicates selling is running out of ammunition. Bullish divergences work somewhat better than bearish ones because bottoms tend to form through exhaustion, which is precisely what volume detects.

The slope is the only thing to read

That last row saves the most money. In a range, the OBV generates false signals by construction, not by bad luck.

Relative volume: how to read the absolute figure

If today's raw volume says nothing, there is a simple way to make it speak: compare it with its own average. Relative volume is the candle's volume divided by the average of the previous twenty. A value of 1 means a perfectly ordinary candle; 3 means three times the usual amount traded.

That figure is comparable across assets and across eras, just as the band width percentile makes volatility comparable. And it solves the underlying problem of this whole subject: it does not matter that Bitcoin trades billions and a small coin trades a few million, because what you look at is how far each one departs from its own habit.

Through that lens, both extremes carry a reading. Very high relative volume on a vertical decline is the typical profile of a capitulation: many people selling at once and at any price, which is how declines tend to end rather than how they tend to begin. Low relative volume sustained for weeks indicates disinterest, and disinterest often precedes large moves, because range compression and volume drought tend to arrive together.

What relative volume does not do is say which way. A spike at three times normal is just as large whether price rises or falls: you have to look at the candle to know which side the pressure was on.

Other ways of reading volume

The OBV is the best known but not the only one, and its alternatives correct exactly its flaw.

Chaikin Money Flow weights by where price closes within the candle's range: closing near the high counts almost all the volume as buying, closing mid-range counts almost none. It is finer than the OBV's yes-or-no.

VWAP answers a different question: at what average price did trading actually occur, weighting each trade by its size. It tells you whether whoever bought today is in profit or in loss, which is different and often more actionable information.

And the volume profile distributes volume across price levels rather than across time, showing where trading concentrated. Those zones tend to act as support and resistance for a mechanical reason: a lot of people hold positions opened there.

Common mistakes

Comparing the OBV of two assets. It makes no sense: each starts from an arbitrary point and accumulates volumes in different units.

Hunting divergences on low timeframes. On 5 or 15-minute candles, closes alternate constantly and the OBV becomes pure noise. Daily or weekly is where it makes sense.

Taking aggregate volume across all exchanges. It adds clean figures to inflated ones and the result is worse than either alone.

Confusing high volume with direction. Enormous volume on a bearish candle is just as high as on a bullish one. Like the ATR, it measures intensity, not direction.

On each asset

On Bitcoin volume is the most reliable in the market and the least affected by manipulation, because it spreads across many large venues. On Ethereum read it beside the ETH/BTC ratio: a volume spike in ETH with a flat ratio usually means market flow rather than specific interest. On XRP the most care is needed, because its volume concentrates in few venues and in sharp news-driven episodes, producing spikes that do not represent sustained participation.

Frequently asked questions