OBV is the simplest volume indicator in common use. Joe Granville introduced it in 1963 with the argument that volume moves before price, so a cumulative volume tally would turn ahead of the price chart. The call is @OBV(close, volume).
How it is computed
OBV_t = OBV_(t-1) + volume_t when close_t > close_(t-1)
OBV_t = OBV_(t-1) - volume_t when close_t < close_(t-1)
OBV_t = OBV_(t-1) when the close is unchanged
The whole day’s volume is assigned to one side based on the direction of the close. A stock that closes up 0.1% credits the full day’s volume as buying, and one that closes down 0.1% debits all of it. There is no weighting by how far price moved.
Notice the signature takes no lookback period. OBV is cumulative from the start of the series, which has a direct consequence: the absolute level is arbitrary. It depends entirely on where your price history begins, and it is not comparable across securities. The one threshold with any meaning is zero:
@OBV(close, volume) > 0
which asks whether cumulative up-volume has exceeded cumulative down-volume since the first bar the strategy loaded. Even that reading shifts if you change the backtest start date, so treat the direction of the series as the signal and the level as an artefact.
Typical uses
| Read | Conventional interpretation |
|---|---|
| OBV rising with price | volume confirms the move |
| OBV flat while price rises | the rally lacks participation |
| OBV falling while price rises | described as bearish divergence |
| OBV rising while price falls | described as accumulation |
Divergence is the traditional reading, and it is also the weakest. Divergences are easy to spot after the fact and appear frequently without any subsequent price move. Counting them honestly on a backtest, including the ones that led nowhere, usually produces a much less impressive picture than the chart examples in textbooks.
Risks and caveats
OBV is unusually sensitive to volume data quality, and that matters on the Indian market. On thinly traded NSE small-caps a single block trade can dominate a week of OBV, and the resulting slope reflects one participant rather than accumulated demand. Reported volume also differs by venue and by whether you count NSE alone or NSE plus BSE, so the same stock can produce different OBV shapes from different feeds.
Corporate actions break the series outright. A 1:5 split multiplies share volume fivefold with no change in rupee turnover, and the cumulative sum carries that discontinuity forever afterwards. Verify your volume history is split-adjusted before using OBV in any rule.
The all-or-nothing assignment is a crude assumption. A day that closes up 0.05% is not economically different from one that closes down 0.05%, yet OBV treats them as opposite events of full size.