Unusual Activity Detector: How Contracts Get Flagged

The Unusual Activity Detector (UA01) answers one question: which option contracts are trading far more than a contract in their position normally trades?

"In their position" is the important part. An option's volume depends heavily on two things: how far its strike is from the current price, and how many days are left to expiry. An at-the-money weekly contract a week from expiry always trades in size. A far strike two months out barely trades at all. Volume only means something when it is compared with the right peers.

The detector works from what the NSE option chain and the NSE bhavcopy actually publish: traded volume, open interest and its change, last and previous price, and implied volatility per contract. The exchange publishes totals per contract, not individual trades or which side initiated them, so the detector does not claim to see sweeps, block orders or the aggressor side. Anything that is missing is shown as missing.

Each contract is compared with comparable contracts from the same symbol's previous expiries: contracts that sat at the same distance from spot, with the same number of days left to expiry. The history comes from NSE bhavcopy.

This matters because the obvious alternative is misleading. Comparing a contract with its own recent sessions sounds reasonable, but a contract's volume grows by orders of magnitude as it approaches expiry and the money. Measured that way, nearly every near-the-money strike looks "hundreds of times above normal" in its final fortnight, which says nothing.

From the comparable contracts the detector derives a volume z-score: how many typical deviations today's volume sits above what those contracts usually traded. The statistics used are robust ones, chosen so that a few past spike days cannot distort the baseline, and they work on a multiplicative scale because option volume is heavy-tailed. The desk also shows the plain multiple (for example 6.2×) next to the z-score.

A contract with too few comparable expiries behind it gets no z-score. It is labelled "no baseline" and can never be ranked above the lowest tier.

The score ranks contracts within one chain. It combines three things measured on the contract itself:

1. Volume against its baseline.

2. Open interest change, because volume that leaves positions behind is different from volume that churns intraday.

3. Premium traded (volume × price), so that a strike trading a few lakh of premium does not outrank one trading hundreds of crore.

OI change and premium are judged against the rest of the same chain on the same session, so "a high score" means "stands out in today's chain", not "crosses a fixed rupee figure" that every large index strike would clear. Near-the-money strikes carry slightly more weight than far wings.

The symbol's ATM IV rank is shown as context but is deliberately not part of the score: it is the same number on every row of a chain and would lift every contract together.

A high score alone never makes a contract unusual. On an ordinary session the largest strikes still score well simply because they are large. Every tier therefore requires the contract's own volume to be above its baseline, and the higher tiers require it to be a statistical outlier:

A liquidity floor removes contracts that traded only a handful of lots, so a far wing trading three lots against a normal of one never appears.

Each row can be opened to see the evidence behind its tier: the multiple of baseline volume, how many comparable expiries stand behind it, open interest and its change, and why the contract did or did not reach the next tier.

Price change against OI change gives the standard four-way read of what happened in a contract:

The desk states this in plain words in the Who acted column (call writers adding, put buyers exiting). The Spot read column then says what that positioning implies for the underlying, which depends on the option type: put writers adding is a bullish read, not a bearish one. If price or OI did not change, the read is neutral rather than forced to a side.

This read is inferred from end-of-interval totals, not from trade-side data. On a strongly trending day most contracts will read the same way, because option prices follow the underlying. Treat it as context for the volume signal, not as a signal of its own.

When a chain source carries no OI change of its own, the detector takes the previous session's open interest and closing price from NSE bhavcopy and says so on the screen. If neither is available, the positioning columns stay blank.

At the 9 November close, the NIFTY 17-Nov 25,000 CE traded about 6.5 lakh lots. Compared with its own earlier sessions, when it was a far strike five weeks from expiry, that was more than a thousand times its usual volume. Compared with contracts at the same distance from spot eight days before previous expiries, it was about 9 times. The detector classed it UNUSUAL, not EXTREME, and the whole NIFTY chain showed 0 EXTREME and 3 UNUSUAL contracts.

On the same session SBIN rose about 2.6%. Its 23-Nov 1,050 CE traded roughly 6 times its baseline with open interest up sharply and the option price about doubling, and was classed EXTREME with the read "call buyers adding".

Read flagged rows as a place to look. Heavy volume on a near-the-money strike can be positions rolling, hedges being placed, or new directional risk. The who-acted read and the strike's position against the call and put walls help tell them apart.

* Common Belief: A contract trading 500 times its own average is extraordinary.

* Reality: Against its own early history almost every near-expiry strike does that. Only a comparison with contracts in the same position is informative.

* Common Belief: A large spike in put volume means a fall is coming.

* Reality: Put volume with rising OI and a falling premium is writers adding, which is usually read as support, not a breakdown.

* Common Belief: Volume alone identifies unusual activity.

* Reality: Volume without OI change is mostly intraday churn; that is why OI change is part of the score.

* Common Belief: Indian option data shows institutional sweeps hitting the ask.

* Reality: The published data carries no trade-level side. The price × OI read is the closest observable proxy.

1. Open OPTIONS → Unusual Activity, or type UA. The desk is part of Professional+.

2. The Monitor board tracks up to 40 index and stock symbols on their front expiry and ranks them by flagged contracts. Each row shows the lead contract, premium traded, the flow read and the session the data is from.

3. Select a symbol to load every liquid contract of its chain below, for any listed expiry. Filter by tier, calls or puts, who acted, or strike, and choose the columns you want.

4. Open a row for the evidence behind its tier.

5. Set alert on a symbol (or Alert on all) to be told when a contract enters a tier. Alerts can be narrowed by side, moneyness, who acted, spot read, premium, volume multiple and OI change, and are delivered through the Alert Center channels.

6. Export CSV writes every scored contract with its provenance: source, session, units and baseline.

During market hours volume is cumulative for the day while the baseline is full sessions, so tiers are conservative until the close: a contract flagged at noon has already out-traded a normal full day. When the market is closed the desk shows the last session and labels it with that session's date.