Volume and open interest: OBV, and the number that is not a price

On Balance Volume

Joseph Granville, 1963, and the ancestor of every volume-based indicator that followed. The rule is one sentence: if the bar closed higher than the last one, add its entire volume to a running total; if it closed lower, subtract its entire volume.

The idea behind it is reasonable. Volume is effort; price is result. If effort is accumulating on the up-closes while price is not yet making progress, someone may be building a position quietly. Granville was trying to see accumulation before it showed in price, which is a good question to ask.

The execution is where it falls down, and it falls down in a way worth understanding properly, because the same flaw appears in half the volume indicators that came after it.

A bar that closes one tick higher and a bar that closes fifty ticks higher are treated identically. Both hand their whole volume to the buyers. Nothing in the calculation knows how hard the close was won, or that both sides were trading the entire time.

So OBV is not measuring buying volume. It is measuring volume, signed by the direction of the close, which is a proxy standing quite far from the thing it stands for. In a choppy session where closes alternate, the running total wanders without describing anything.

The honest summary: OBV was an excellent question asked with the only data available in 1963. In futures, the data needed to ask it properly does exist now — which trades crossed the spread, and in which direction — and that is delta and the footprint, both written earlier in this path. If you understand those, OBV is history rather than a tool, and it is worth knowing as history because its logic is still embedded in indicators sold today.

The same caution extends to its descendants — money flow measures, accumulation and distribution lines. They refine how much of the bar's volume to assign to each side, usually by where the close sat within the range, but they are still assigning. Refining a guess does not turn it into a measurement.

Open interest

Now the exception to everything the first page of this section said.

Open interest is not a calculation performed on your chart. It is a count, made by the exchange, of how many contracts are currently open — positions entered into and not yet closed out. It arrives from outside. Nothing you could compute from open, high, low, close and volume would give it to you.

The distinction from volume is the whole point and is constantly muddled:

The conventional way to read the pair is worth knowing, with the caveat attached rather than left off: price rising while open interest rises is usually described as new positions being built into the move, and price rising while open interest falls as existing shorts closing out. The caveat is that this is an aggregate across every participant in the contract, and an aggregate cannot tell you about the particular participant moving price in front of you right now.

Two things about it that catch people out

• It is not real time. The exchange publishes open interest after the session, not tick by tick. Watching it intraday for a reaction to something that just happened is waiting for a number that is not coming until tomorrow.
• It is per contract month. Which makes it the clearest possible view of a problem this path has already covered from the other side: when open interest starts draining out of the front month and appearing in the next one, the roll is under way, and the liquidity is moving whether or not you move with it. See the roll over.

What neither of them tells you

• Who. Open interest counts contracts, not participants, and has no idea whether a position is a speculation, a hedge against something in the physical world, or one leg of a spread.
• Where the positions were opened. A count of open contracts carries no price with it, so «the crowd is offside» is not a conclusion you can draw from this number alone.
• Anything about the next hour. One is a running total of the past; the other is yesterday's census.