On 10 Jun 2026 the Nifty closed at 23,214.95. In module A19 the same chain gave a small positive gamma total under one sign rule, a large negative one under another, and small positive ones when the sign came from NSE's participant file. That file is the only public data on who holds what, and it is by group alone. So I ask a harder question here: does using it make the answer firm? Pick the group that acts as the dealer and the total changes sign again. Pick another date and it changes sign again. A positioning model built on public data is a useful sketch. This lesson builds one, runs it on dated chains, and lists exactly where it stops being evidence.

3By the end you can

  1. Turn NSE participant wise open interest into a signed gamma total, step by step.
  2. Show how the total changes with the dealing group, the date and the units.
  3. List the five assumptions inside the model and what a fair test of it would need.
The RupeeCase terminal on sample data: Compare strategies: how do I set up to four side by side?
Watch in the terminal: how do I set up to four side by side? A short tutorial on sample data.

The idea, plainly

What the participant file holds

NSE publishes a daily file of open interest by participant. It has four groups: Client, DII, FII and Pro, the file's label for proprietary accounts of trading members. For each group it gives long and short contracts in index futures, index call options and index put options. That is all. There is no strike. There is no expiry. All index contracts are pooled: Nifty, Bank Nifty and the others sit in the same rows. It is end of day, and it counts contracts, while the option chain counts units.

The groups net to zero, so any dealer you pick puts another group on the other side

Net option positions by participant group on 10 Jun 2026, in contracts, all index options pooled: Client net short puts and FII net long puts are the largest, and the four groups net to zero in each productClient, calls 2,15,246; Client, puts −7,67,837; FII, calls −2,77,427; FII, puts 5,30,987; Pro, calls 54,028; Pro, puts 2,12,473; DII, calls 8,154; DII, puts 24,377.Net contracts, long less short; L is lakhClient, calls+2.15LClient, puts−7.68LFII, calls−2.77LFII, puts+5.31LPro, calls+0.54LPro, puts+2.12LDII, calls+0.08LDII, puts+0.24L
NSE participant wise open interest for 10 Jun 2026, long less short, for index calls and puts. The file pools all index contracts and counts contracts, not units. Source: NSE end of day data, computed by RupeeCase. As of 10 Jun 2026.
Read one more paragraph, 83 words

Within each product the groups net to zero. Every long has a short somewhere, so whatever you assume about one group puts someone else on the other side. Look at the chart below for the size of it on 10 Jun 2026. One group, Client, was net short puts by a wide margin, and FII was net long them by a smaller one. DII held few options at all next to the others, which will matter when DII is chosen as the dealer.

Convention C, step by step

Convention C of module A19 turns the file into a sign for every line of the chain. The steps are these.

  1. Pick the group that acts as the dealer. This is a choice, not a fact.
  2. For calls, take the group's long less short contracts and divide by all long calls of all groups. That is the dealer's net share of call open interest. Do the same for puts.
  3. Give every call line of the Nifty chain that share as its sign, and every put line its own share.
  4. Compute gamma exposure for each line as before: gamma times open interest in units times spot squared times 0.01, in rupees crore.
  5. Add. The total is positive if the chosen group is net long gamma on these assumptions.
The maths: A, B and C as one family

Let each line carry a sign s. Gamma exposure of a line is s times a fixed positive size. Then the three conventions differ only in s.

A: s = +1 calls, −1 puts
B: s = −1 calls, −1 puts
C: s = share of the group
(a fraction between −1 and 1)

A and B give every line its full size. C scales each line by a fraction, the group's net share of that product. When the chosen group holds only a small net share, as a proprietary group usually does next to all long open interest, every line shrinks and the total is small and its sign is decided by small differences. That is why C totals are a few thousand crore while B totals are tens of thousands.

Why net, and not gross

A group can hold large long and large short positions in the same product, and they cancel. A market maker who quotes both sides carries big gross numbers and a small net one. Gamma exposure follows the net, since a long and a short of equal size give no hedge to trade. That is why the model uses long less short and not the long side alone. It is also why a group with huge gross open interest can add almost nothing to the total.

Read one more paragraph, 61 words

The net share is also a ratio to the whole product, so it is small unless a group holds a large part of all open interest on one side. A group that holds a tenth of long calls and is net long by half of that has a share of five hundredths. Every line of the chain is multiplied by it.

2 terms this section turns on

2 terms this section turns on2 terms this section turns on. Pro: The file's label for the proprietary accounts of trading members. It is a type of account, not a role such as market maker. Net share: A group's long less short contracts in a product, divided by all long contracts of all groups in that product.ProThe file's label for the proprietary accountsof trading members. It is a type of account,not a role such as market maker.Net shareA group's long less short contracts in aproduct, divided by all long contracts of allgroups in that product.
From the module glossary; every term is in the Learn glossary too. Source: Module 15.13, An Indian positioning model and its limits.

Five assumptions inside it

  • The group is the dealer. A label in a file says who the account belongs to, not what it does. A proprietary group holds directional books, arbitrage books and market making together.
  • The net position is spread over strikes in proportion to open interest. The file has no strike, so the model has to pour the net share evenly over every line.
  • The pool is the chain. The shares come from all index contracts, the chain is Nifty alone. Bank Nifty, with different positions, is inside the shares.
  • The hedge is daily and in the future. A dealer may hedge with other options, across the day, or not at all.
  • Volatility sticks to the strike. The profile reprices with each strike's volatility held fixed, which a moving market does not do.

Two checks on units. NSE open interest on the chain is in units, which are contracts times the lot of 65. The participant file is in contracts. Put them side by side only after dividing by the lot.

What would settle it

Positions by account, strike and expiry, with the side. Only the exchange and its clearing corporation see that. SEBI's own studies of traders use account level records that are not public. Until a file like that exists, a positioning model is a model, and its outputs are readings under assumptions.

The RupeeCase terminal on sample data: Model history or your portfolio: which numbers am I looking at?
Watch in the terminal: which numbers am I looking at? A short tutorial on sample data.

Worked example: four groups, five dates, 10 Jun 2026 first

On 10 Jun 2026 the 23,500 call of the 30 Jun series had 44,81,555 units of open interest. At the lot of 65 that is 68,947 contracts. The chain held 6,57,39,570 call units and 6,85,47,860 put units in that series, a put to call ratio of 1.04. Now run convention C four times, once for each group as the dealer.

Choose Client as the dealer and the total changes sign

Total gamma exposure of the Nifty chain on 10 Jun 2026 under convention C with each group as the dealer: positive for Pro, FII, DII and the two together, negative when Client is the dealerPro as dealers 2,047; FII as dealers 1,914; Pro and FII together 3,960; DII as dealers 250; Client as dealers −4,210.Total GEX, convention C, Rs crore for a 1% movePro as dealers+2,047FII as dealers+1,914Pro and FIItogether+3,960DII as dealers+250Client as dealers−4,210
The 16 Jun and 30 Jun series on 10 Jun 2026, expiries up to forty five days, rupees crore for a one per cent move. Only the choice of dealing group changes between bars. Source: NSE end of day data and NSE participant wise open interest, computed by RupeeCase. As of 10 Jun 2026.
Read one more paragraph, 93 words

Four of five readings are positive. The fifth is negative, and it treats the group that holds most of the open interest as the hedger. DII holds few options, so its reading sits near zero whatever the sign. FII and Pro give small positive totals of similar size, though they hold opposite positions in calls. Client is the mirror of the other three groups taken together, because the groups net to zero, so its total has the opposite tone. The file cannot say which reading describes what happened in the market that day.

The same rule on five dates

Fix the rule: Pro as the dealer. Run it on the five sessions for which these lessons have full Nifty chains.

One rule, five dates, and the sign is not stable

Total gamma exposure under convention C with Pro as dealers on five sessions: negative on the two February sessions and positive on the three in June1 Feb 2026, Budget Day −1,204; 2 Feb 2026 −1,381; 10 Jun 2026 2,047; 16 Jun 2026 179; 29 Jun 2026 1,993.Total GEX, convention C with Pro as dealers, Rs crore1 Feb 2026,Budget Day−1,2042 Feb 2026−1,38110 Jun 2026+2,04716 Jun 2026+17929 Jun 2026+1,993
Convention C with Pro as the dealers, nearest series up to forty five days, rupees crore for a one per cent move. The sign follows the date. Source: NSE end of day data and NSE participant wise open interest, computed by RupeeCase. As of 29 Jun 2026.

Negative on two sessions and positive on three, with the size on 16 Jun close to zero. Five dates are far too few to say anything about how often it flips. They are enough to show the sign is not a fixed feature of the market that one can quote.

Which choice matters most

The model has more than one dial. Besides the group and the date there is the expiry cut off, which drops series further out than some number of days, and the source of each leg's volatility. Here is each dial turned on its own, from one base: Pro as the dealers on 10 Jun 2026, expiries up to forty five days, each leg at its own volatility.

The group and the date move the answer far more than the expiry cut off or the volatility source

Total gamma exposure under convention C, one choice changed at a time from a base of Pro as dealers on 10 Jun 2026: the expiry cut off and the volatility source change its size but not its sign, while the dealing group and the date change its signBase: Pro, 45 days 2,047; Expiries to 7 days 1,462; Expiries to 60 days 2,126; Volatility of OTM leg 2,044; FII as dealers 1,914; Client as dealers −4,210; Date: 1 Feb, Pro −1,204.Total GEX, convention C, Rs crore for a 1% moveBase: Pro, 45days+2,047Expiries to 7days+1,462Expiries to 60days+2,126Volatility ofOTM leg+2,044FII as dealers+1,914Client asdealers−4,210Date: 1 Feb,Pro−1,204
Each bar changes one input from the base bar. The two choices that flip the sign are the dealing group and the date. The expiry cut off and the leg volatility change the total by a fraction of its size. Source: NSE end of day data and NSE participant wise open interest, computed by RupeeCase. As of 10 Jun 2026.
Read one more paragraph, 82 words

The expiry cut off moves the total by up to about a third, and the volatility source hardly at all. Neither changes its sign. The other two flip it. The dials that flip it are the two that no file settles. The group is a label you choose and the date is the day you happened to look. That is the shape of a fragile result: the inputs you can measure barely matter, and the inputs you cannot measure decide the answer.

Where the bars sit

Here are the strikes of 10 Jun under convention C with Pro as the dealer. The call wall and the put wall are the strikes with the largest call and put lines.

Under convention C with Pro as dealers every bar is small and positive

Net gamma exposure by strike under convention C with Pro as the dealers on 10 Jun 2026: small positive bars at every strike, because the group is net long both products by a small shareNet gamma exposure by strike under convention C (participant informed (NSE participant wise open interest)), in ₹ crore per 1% move of spot. 37 bars positive, 0 negative; the largest bar is 261 at 23,225. Each bar adds up 2 neighbouring strikes. Spot 23,215.Convention C: participant informed (NSEparticipant wise open interest)₹ crore per 1% move of spotPositive net gammaNegative net gamma05010015020025022,00023,00024,000Strike priceSpot 23,215Call wall 23,500Put wall 23,000
Net gamma exposure by strike on 10 Jun 2026 for the 16 Jun and 30 Jun series, expiries up to forty five days. Rupees crore for a one per cent move, with the group's net share as each line's sign. Source: NSE end of day data, computed by RupeeCase. As of 10 Jun 2026.
Read one more paragraph, 58 words

Compare the scale with the A19 charts. Every line is a fraction of its full size, so the bars are short and all point the same way, because the group is net long both calls and puts by a small share. A flat picture tells you the group's net share is small, not that the market is calm.

What the model can still do

Used with care it is a consistent ruler. Fix the rule, then compute the same figure on each session. Differences between sessions are then differences in open interest, volatility and time, and not in the rule. That is a description, and the lab beside this lesson does exactly that. What it cannot do is name the sign of anyone's position, tell you the flow that will follow, or say that a day will be quiet or wild. For that the missing input is the one the exchange keeps.

Read one more paragraph, 41 words

The participant file is also published after the close and describes the close. By the time it is read, the next session is already being priced. A model that needs the file to know the sign is a model of yesterday.

The chain of assumptions as a picture

Every arrow in the chain carries an assumption

Convention C as a chain of steps from the participant file through the choice of dealing group and net shares, joined with the Nifty chain lines, into a signed total that is read as a market viewDiagram with 6 boxes. Participant file to Choose the dealing group; Choose the dealing group to Net share of each product; Net share of each product to Signed total and profile; Nifty chain lines to Signed total and profile; Signed total and profile to Read as a market view.Participant filecontracts, no strike, allindicesChoose the dealing groupa label, not a roleNet share of eachproductone number for calls,one for putsNifty chain linesunits, strike bystrikeSigned total and profilesticky strike volatilityRead as a market vieweach step above is a guess
The file has no strikes, so the net share is spread over the chain. The chosen group is a label. The total is only as firm as its weakest step. Source: RupeeCase illustration of convention C. As of 17 Oct 2026.
Appendix: a test I would accept

Write the rule before looking at results: the convention, the group, the expiry cut off and the volatility source. Then fix one target, such as the next session's absolute move divided by the move that India VIX implied, and one baseline, such as using only India VIX. Collect every available session of chains and participant files, not the vivid ones. Fit nothing on the later half and test on it, as in module A29. Count every variant you tried. Four groups, four expiry cut offs and three volatility sources make 48 readings, and the best of them will look good by chance, which module I34 explains. Report the result against the baseline, with the sample size. This lesson's five dates cannot do any of that, which is why it claims nothing.

The common mistake

The claim: Pro are the market makers, so the Pro column of the participant file gives dealer gamma, and positive means a quiet day. The label says a group of accounts, not a role. Pro holds directional and arbitrage books beside market making. On the five sessions here the Pro based total is negative on two and positive on three, and the sign changes again when the dealing group is Client. The file is pooled across index contracts and has no strike. A signed total from it is a reading under five assumptions. Ask which group was chosen and why, and how many readings were tried before this one was shown.

Key takeaways

  • The participant file has no strike or expiry, and pools every index contract.
  • Convention C turns the file into a sign by choosing a dealing group.
  • On 10 Jun 2026 four of five dealer choices gave positive totals.
  • With Pro as dealers the sign differed across five dated sessions.
  • Chain open interest is in units; the participant file counts contracts.
  • A positioning model is a reading under assumptions, not a measurement.

Two terms to hold on to from 15.13

Two terms to hold on to from 15.13Two terms to hold on to from 15.13. Participant wise open interest: NSE's daily count of long and short contracts by Client, DII, FII and Pro, for index futures and index options, with no strike and no expiry. Dealing group: The group a model treats as the hedger. It is a choice made by the analyst, not a field in any file.Participant wise open interestNSE's daily count of long and short contractsby Client, DII, FII and Pro, for index futuresand index options, with no strike and no...Dealing groupThe group a model treats as the hedger. It isa choice made by the analyst, not a field inany file.
From the module glossary. Source: Module 15.13, An Indian positioning model and its limits.

Before you trade. Derivatives trading carries a high risk of loss. SEBI's study "Profitability of Individual Traders in the Equity Derivatives Segment (FY25 to FY26)" (August 2026) found that 87.7% of individual traders in equity F&O made net losses in FY26, with aggregate net losses of Rs 91,685 crore. Read the SEBI study