Open Interest (OI) & The Four Phases

Volume is a count of contracts traded during a session. If five thousand lots of the NIFTY 24,500 call change hands between 09:15 and 15:30, the volume on that strike is five thousand lots. It makes no difference whether those trades opened new positions or closed old ones — every trade adds to the count. The number starts again from zero at the beginning of the next session. Volume is a flow: a measure of activity over a window, and nothing more. A high figure tells you a strike was busy. It does not tell you whether anybody stayed.

Open interest counts contracts that are still open — not squared off, not exercised, not expired. It is a stock, not a flow. It does not reset overnight; it carries forward until the position is closed or the series expires. Open interest rises by one lot only when a fresh buyer and a fresh seller together create a contract that did not exist before. It falls by one lot only when an existing buyer and an existing seller both close. Every unit of open interest has exactly one long and exactly one short attached to it, always.

The pair is far more informative than either number alone. Suppose NIFTY futures traded 3.1 lakh lots in a session and open interest finished at 2,34,000 lots against 2,10,000 the previous evening. Volume of 3.1 lakh against a net addition of 24,000 lots says most of the day was intraday churn, but 24,000 lots of genuinely fresh commitment did stay on the table. Now suppose the identical volume ended with open interest unchanged at 2,10,000. Same activity, opposite conclusion: everything opened during the day was also closed during the day, and nobody is carrying anything new into tomorrow.

Every trade has a buy side and a sell side, and each side is either opening a new position or closing an existing one. That gives four possible combinations, and only two of them move open interest at all. This is why a violently busy session can leave the number untouched. It is also why the change in open interest, rather than its level, is the figure worth reading — the level tells you how crowded a contract is, but the change tells you what happened today.

Follow one contract through three sessions. In the first, trader A buys one lot of NIFTY futures from trader B, both of them opening. Volume is one lot and open interest is one lot. In the second session A sells that lot to trader C, who is opening a fresh long. A is out, C is in, and the contract still exists: volume one lot, open interest still one lot. In the third session B buys the lot back from C, and both are closing. Volume is one lot again and open interest drops to zero. Three sessions, three lots of volume, and open interest went 1, then 1, then 0.

The exchange does not publish which of the four combinations occurred on any individual trade. What it publishes is the net position at the end — today against yesterday. Everything that follows in this module is therefore an inference drawn from two published numbers, not a disclosed fact about who did what. That distinction is worth holding on to, because the four phase labels are usually presented with a confidence the underlying data does not support.

Take two directions of travel for price — up or down — and two for open interest — expanding or contracting. That gives a two-by-two grid with four boxes, and each box has a name. The logic behind all four is identical: expanding open interest means contracts are being created, which requires fresh participants on both sides; contracting open interest means contracts are being extinguished, which requires existing participants on both sides to leave. Price then tells you which side was the aggressor.

Price up with open interest up is called long buildup. New contracts are being created while the price climbs, so fresh money is arriving in support of the move rather than old positions being unwound. Price down with open interest up is short buildup: new contracts again, but the price is being pushed down, so the fresh participation sits on the sell side. In both cases the market is getting more crowded, and the size of the position the market is carrying overnight has grown.

The other two boxes are exits. Price down with open interest down is long unwinding — existing longs are closing out and the contracts vanish, so the fall is being driven by people leaving rather than by fresh sellers arriving. Price up with open interest down is short covering: existing shorts are buying back, contracts vanish, and the rise is driven by exits. Short covering moves can be sharp precisely because a short that is buying back is not choosing to buy at a good price, it is closing an obligation. But sharp is not the same as durable, and no fresh money has entered on either of these two.

Below is an illustrative run of five NIFTY futures sessions. Only two columns matter: where the contract closed and where open interest finished against the previous evening. Work down the table and each of the four phases appears once. Notice that the index visits roughly the same level twice — around 24,400 in session two and session four — and the reading is completely different on each visit, because the open interest behaviour was different.

Session two adds 24,000 lots on a 130-point rise. That is a long buildup: fresh participation arriving with the move. Session three rises again, but 6,000 lots come off, so the second leg of the rally was existing shorts closing rather than new buyers arriving. Session four falls 115 points and adds 13,000 lots, which is short buildup — the fall was driven by new selling. Session five falls again and drops 22,000 lots, so that leg is longs exiting rather than shorts pressing.

The important discipline here is that you read the change, never the level. Open interest of 2,34,000 lots is neither high nor low on its own; it only becomes meaningful against 2,10,000 the evening before, and against the size of the changes over the previous ten sessions. A 2 per cent change on a quiet session and a 2 per cent change on a policy-announcement session are not the same event, and comparing today only against zero throws that context away.

No, and this is where the framework is most often stretched past what it can carry. In futures there is one contract and one open interest series, so a rise in open interest is unambiguous about the fact that contracts were created. In options there is a separate open interest figure at every strike, on both the call and the put side, and every one of those numbers has the same limitation: it counts pairs. A rise of 40,000 lots in call open interest at the 24,700 strike means 40,000 new call contracts exist. It does not tell you whether the aggressor was a writer or a buyer.

That matters because the shorthand you see everywhere — highest call open interest marks resistance, highest put open interest marks support — rests entirely on the assumption that the open interest at those strikes was created by writers. Writers do have a reason to defend a strike, and that assumption is often reasonable on far strikes in an index chain. But it is an assumption, not a reading. The exact same open interest figure would appear if a large number of buyers had paid up for that strike expecting the index to travel through it.

Multi-leg positions muddy it further. A trader putting on a spread adds open interest at two strikes simultaneously while carrying a modest directional view, and a hedged institutional position adds open interest while carrying no directional opinion at all. The option chain module covers how to read the columns; the point here is narrower. Futures open interest is one number about one contract. Option open interest is a distribution across strikes, and a distribution answers a different question.

Start with the biggest one: open interest carries no direction. Every open contract has exactly one long and exactly one short, so open interest counts pairs. A statement such as "open interest shows that institutions are short" cannot be read off the number, because the number is symmetric by construction. Participant-wise positioning is published separately by the exchange as an end-of-day file, and that is a different dataset with a different release schedule. The open interest column on your screen is not it.

It also carries no identity and no motive. Retail, proprietary desks, foreign investors and pure hedgers are all pooled into one figure. A textile exporter hedging currency exposure and a day trader taking a punt contribute identically. And because open interest is a count of contracts, not of rupees, it tells you nothing about the price at which those positions were established — so any claim about the average cost of the crowd, or where the crowd is trapped, is being read into the data rather than out of it.

The third gap is time horizon. A contract that was opened this morning and one that has been carried for four weeks look the same in the open interest figure. The fourth is the most expensive in practice: open interest is reported per series, and as a series approaches expiry, open interest falls because positions are being rolled into the next series. That decline is a calendar event, not a change of view. Reading it as long unwinding is one of the most common errors in open interest analysis, and it recurs every single expiry cycle.

Finally, and most importantly, none of the four phases is a forecast. A large short buildup is consistent with a further fall and equally consistent with a violent squeeze higher — the squeeze is what a crowded short book makes possible in the first place. The framework describes what has already happened to positioning. It does not contain information about the next session, and no circulated percentage claiming otherwise discloses the instrument, the period or the sample it was measured on.

Open interest works best as a filter on a read you already have, not as a signal generator on its own. If price action has given you a level — a breakout from a multi-candle consolidation, a retest of a broken support — then the open interest behaviour around that level either supports the idea that fresh money is behind the move or it does not. That is a genuine input. Starting from the open interest number and reasoning backwards to a view is how traders end up with confident conclusions built on a symmetric, anonymous, direction-free statistic.

The second discipline is to always check the next series before concluding anything about a fall in open interest. If the near series lost 40,000 lots and the next series gained roughly 40,000, the position has moved, not disappeared. This single check removes a large share of the false unwinding readings that appear every expiry cycle, and it takes about ten seconds on the exchange site.

The third is to build your own record rather than borrow somebody else’s. Log the close, the change in open interest and the phase label for one instrument you actually follow, every session, for a couple of months. After forty sessions you will have a table that tells you how often each phase was followed by continuation in that instrument, over that period, at that volatility regime. That is worth considerably more than any circulated hit rate, because you know exactly what it was measured on. Pair it with the put-call ratio and max pain modules, which read the same option chain for different questions.