What open interest counts
Two trades of the same size print seconds apart. After the first, open interest rises. After the second, it does not move at all. Later, a burst of trading goes through and the number falls while volume climbs.
Open interest is not a measure of activity. It is a count of contracts that have been opened and not yet closed, and whether a given trade changes it depends entirely on what the two parties were each doing with their existing positions. Once that is clear, every one of the confusing cases resolves into arithmetic.
A stock, not a flow
Volume is a flow: it accumulates over a period and resets. Open interest is a stock: a level that exists at an instant and is carried forward.
Concretely, open interest is the quantity of contracts currently outstanding. If nobody trades for an hour, volume for that hour is zero and open interest is exactly what it was. If a great deal of trading happens and every trade is one participant handing a position to another, volume is large and open interest has not moved.
There is a counting convention to settle first, because venues differ. A contract necessarily has a long side and a short side. Some venues count the contract once; some count both sides and publish a figure twice as large. Neither is wrong and the two are not comparable, which is enough on its own to make a figure from one venue meaningless against a figure from another.
The four combinations
Every trade pairs one buyer with one seller, and each of them is either establishing a position or retiring one. That gives four cases, and only two of them move the number.
Both opening. A new long meets a new short. Contracts that did not exist now exist, and open interest rises by the traded quantity.
Both closing. An existing long sells to an existing short who is buying back. Contracts that existed are extinguished, and open interest falls by the traded quantity.
One opening, one closing. The position moves from one holder to another. The number of outstanding contracts is unchanged. This is a transfer, and it is the most common case in an active instrument.
The important structural point: long and short quantities are equal by construction. Every contract has both sides. There is no state of the world in which open interest expresses more longs than shorts, because the two are the same number viewed from opposite ends.
Why it moves when you are not looking
Several mechanisms retire contracts without anyone deciding to trade.
A liquidation closes a position with an order the account holder did not send, so open interest falls. Where the venue’s shortfall machinery closes positions on the other side as well, that reduces it further, and both legs of the reduction were involuntary.
Contracts with an expiry are retired at settlement, taking the instrument’s open interest to zero by rule rather than by trading. And the published figure is a measurement: it is taken on some cadence, so a level you read is the level at the last computation, in the same way that a depth display is a snapshot rather than the live book.
Contracts versus notional
A second convention worth separating. Open interest can be published as a count of contracts or as a notional value — contracts multiplied by a price.
The notional version moves when the price moves, with no trading and no change in the number of outstanding contracts. It is arithmetic on two inputs and it is routinely read as though only one of them varies. The denomination of the contract decides which of the two figures is even natural to quote, since an inverse contract’s size is already defined in the quote currency.
What it does not tell you
Three things, stated plainly because each is a common reading.
It is not liquidity. Open interest describes positions held; depth describes orders resting. Those are separate dimensions and a large outstanding position count is entirely compatible with a thin book, because holding a contract puts nothing in the book.
It is not volume. The two are computed from the same trades and answer different questions, which is why they diverge routinely rather than exceptionally.
And it is not a direction. This site attaches no predictive interpretation to open interest, rising or falling, in isolation or alongside anything else. What is mechanically true is only that a rise means contracts were created, a fall means contracts were retired, and both sides of every one of them exist.
The one place it is directly load-bearing is funding: since a funding payment is a transfer between the holders of outstanding positions, open interest is the quantity across which that transfer applies.
The mechanism
THE MECHANISM — a contract count
· Buyer opening, seller opening
→ contracts created. Open interest
RISES by the quantity.
· Buyer closing, seller closing
→ contracts retired. Open interest
FALLS by the quantity.
· One opening, one closing
→ transfer. Volume records the
trade; open interest does not.
· Nobody trades for an hour
→ volume zero, open interest
unchanged. A stock, not a flow.
· A liquidation or a shortfall close
→ open interest falls on orders
nobody chose to send.
· Published as notional rather than
contracts
→ the figure moves on PRICE ALONE,
with no trading and no change in
contracts.
· Read as long-versus-short positioning
→ NOT AVAILABLE. The two sides are
equal by construction.
· One-sided versus two-sided counting,
measurement cadence, and contract
versus notional convention
→ VENUE-SPECIFIC. Figures from two
venues are not comparable.Worked example
Illustrative figures throughout, synthetic and round, describing no real venue. Assume one-sided counting and contracts of 1 unit each.
Start with open interest at 100 contracts.
Trade one. Participant A buys 10 to open a new long; participant B sells 10 to open a new short. Both opening, so 10 contracts are created. Open interest is 110. Volume for the session is 10.
Trade two. A sells 4, closing part of the long; participant C buys 4, opening a new long. One closing, one opening. Nothing is created or retired. Open interest is still 110. Volume is 14.
Trade three. B buys 6 to close part of the short; A sells 6 to close more of the long. Both closing, so 6 contracts are retired. Open interest is 104. Volume is 20.
So across three trades of 10, 4 and 6 units, open interest went up by 10, then nowhere, then down by 6 — and the trades were indistinguishable in size and appearance on the tape, which records the trade and not the intent behind either side of it.
The notional restatement. At a price of 40,000, those 104 contracts are a notional 4,160,000. Suppose the price then rises to 44,000 with no further trading whatsoever. Contract open interest is 104, exactly as before. Notional open interest is 4,576,000, an increase of 10%. Nothing was opened.
The failure mode
The characteristic error is treating open interest as an observation about participants when it is an identity about contracts. Every one of its surprising properties follows from that.
Three of them, all documented behaviour rather than reporting error. A trade that looks identical to the previous one can move the figure in either direction or not at all, and the deciding information — who was opening and who was closing — is never published per trade. The figure can fall sharply with no voluntary selling anywhere, because forced closes retire contracts on both sides. And a notional figure conflates price with participation so completely that a session with no trades at all can show a double-digit change in it. Before reading anything into a movement, the convention, the cadence and the unit have to be established for the specific venue publishing it, and that is a question about the publisher rather than about the market.