How institutions trade in India: what shows up on the NSE tape

Mutual funds, insurers, foreign funds and trading firms move orders worth hundreds of crores. They cannot buy the way a retail trader does, all at once at the market price, without moving it against themselves. So they use different tools, and those tools leave marks on the tape that anyone can learn to read. Here is how institutional orders are executed, and what they looked like in our records over the last few weeks.

How a large order actually gets done

• Split over time, judged against VWAP. A large order is sliced into many small ones and fed in through the session by an execution algorithm. The desk is usually judged against the day's volume-weighted average price, so it tries to trade in step with the market's own volume (VWAP explained).
• Block deals in their own windows. Deals of at least ₹25 crore can be done in two short block windows, 8:45 to 9:00 (priced within 3 % of the previous close) and 2:05 to 2:20 pm (within 3 % of the VWAP between 1:45 and 2:00). Those are the rules SEBI set in October 2025, in force since 7 December 2025.
• Pre-arranged trades. Two parties who have already agreed a price can cross a large quantity at once, and the price barely moves.
• Direct access and co-location. Institutions and trading firms route orders straight into the exchange through their broker's infrastructure, and some keep servers in the exchange's own data centre to cut the delay.

What it looked like on our tape

Block and bulk deals. Between 19 August and 11 September 2026 (18 sessions), about ₹16,600 crore changed hands in block deals, and 3,310 bulk-deal disclosures were published. The largest involved institutions on at least one side:

From the exchange's public block and bulk deal disclosures. Where the other side is an individual, the name is left out.

Trades that do not move the price. From 1 to 11 September, about a third of big cash-market prints of ₹3 crore or more, 2,179 of 6,299, left the price exactly where it was: pre-arranged crosses, not someone pushing. They happened all day and were heaviest after 3 pm (₹5,641 crore). Only 169 big cash prints (₹1,084 crore) fell inside the two block windows, so most institutional size reaches the ordinary tape. More on this in big prints on NSE.

A rollover. On 3 September at 10:47, ₹357 crore traded in HDFCBANK's September future and ₹359 crore in the October future in the same minute, neither moving the price: a position moved from one month to the next.

An index rebalance. On 31 August, the day index changes took effect, FIIs sold a net ₹7,986 crore in the cash market and the official closing prices of F&O stocks landed an average 1.5 % away from their last continuous trade, because rebalancing orders go into the closing auction (how the closing auction works).

A steady buyer. On 11 September, trades lifting the offer outweighed trades hitting the bid in NIFTY futures in every hour from 10:00 to the close, while NIFTY climbed from a gap down into the previous days' range: the pattern of a large order being fed in over the day rather than a single push.

What a retail trader can read from this

• Size that does not move price is usually a deal, not a view. Check whether the price moved before reading anything into a large print.
• Steady one-sided flow matters more than one big print. Of 4,637 big cash prints we followed, price 30 minutes later went their way only 44 % of the time; a whole session of net buying tells you more than a single trade.
• Know the calendar. Rebalance days, expiry days and rollover weeks carry institutional flows that are about the calendar, not about the stock.
• FII and DII net figures are the day's summary, not the push. See today's FII and DII data.

Sources

• Block deal rules: SEBI circular on the block deal framework, 8 October 2025, as reported by Business Standard.
• Block and bulk deals: the exchange's public disclosures, as recorded by TBTflow, 19 August to 11 September 2026.
• Big prints, crosses and futures flow: TBTflow's own NSE tick and order-book record.