An ETF trades at Rs 101.2 while its indicative NAV is Rs 100.0. A creation unit is 50,000 units, creating costs 0.3% and trading the basket costs 0.2%. Is creation arbitrage profitable, and what closes the premium?

Yes: the 1.2% premium beats the 0.5% cost of creating, leaving Rs 0.70 a unit, or about Rs 35,000 on one creation unit of 50,000. The participant buys the basket, delivers it for new units and sells them at Rs 101.2. That selling is what closes the premium: arbitrage keeps pushing until the price sits within about 0.5% of NAV, the band where creating no longer pays.

Think of a sweet shop that sells gift boxes of assorted sweets. If the boxed version sells for more than the loose sweets inside plus the cost of the box and the packing, anyone who can pack boxes buys loose sweets, packs them and sells boxes. An authorised participant does the same: it buys the basket of shares at NAV, hands it to the fund for new ETF units, and sells those units at the higher market price. Here the market price is Rs 101.2 against a basket worth Rs 100.00, a 1.2% premium. The costs are 0.3% creation fee and 0.2% to trade the basket, Rs 0.50 a unit, so Rs 0.70 is left.

Size it. One creation unit needs a basket worth about Rs 50,00,000, roughly Rs 50 lakh, and returns Rs 35,000 if prices hold still while the trade is done. That is a 0.7% return on capital tied up for hours, which is why participants act quickly and why premiums of this size rarely last. The participant also needs to sell the 50,000 new units without pushing the ETF price down too far itself; if the ETF's own order book is thin, selling the units costs part of the edge.

The arbitrage itself. Every creation adds new ETF units to the supply and they are sold into the market, so the ETF price falls towards NAV. The selling stops when the premium no longer covers the cost of creating, at about Rs 100.50; on the other side, if the ETF trades below about Rs 99.50, participants buy units and redeem them for the basket, pushing the price up. So an ETF does not trade exactly at NAV; it trades inside a band whose width is set by the cost of creating and redeeming. Cheaper, more liquid baskets mean a narrower band.

When the indicative NAVA live estimate of the value of the basket behind each ETF unit, computed from the latest prices of its holdings. It is only as good as those prices. is stale. If some of the basket is not trading, because shares are halted or the underlying market is shut, the ETF price may be the more accurate estimate of value, and the premium is a forecast, not a free profit. For this domestic large-cap fund on a busy morning that is unlikely, but it is the first thing to check. The second is timing risk: if the basket's price rises while the participant is buying it, the edge can vanish before the units are sold, so participants hedge with index futures while they work the order.