FII / DII Activity: methodology

Cash flows. Provisional FII/FPI and DII buying, selling and net figures in ₹ crore, for NSE, BSE and MSEI combined, as the exchanges publish them each evening. FII figures start on 4 April 2006 and DII figures on 16 April 2007. Daily figures from the exchange; earlier history and any day a daily update missed come from Moneycontrol's FII & DII archive, which carries the same provisional figures. That archive lacks about 1% of trading days since 2012 (some special Saturday sessions among them); those days are absent here. These are not NSDL's settlement-date FPI figures, which differ.

NIFTY 50. Daily closes. Before 21 February 2012, and on the few later days the price data lacks, the close comes from the same archive as the flows (the two agree exactly where both exist). Monthly and yearly returns run from the last close of the previous period.

Monthly & yearly. Sums of the daily net figures. A streak is consecutive sessions with FII net buying (or net selling). "DII net ÷ FII net sales" is the year's DII net figure divided by the FII net sales (1.0× = DIIs bought exactly what FIIs sold, net), shown only for years in which FIIs were net sellers.

Derivatives positioning. NSE's participant-wise open interest in index futures and index options (number of contracts, from 2 January 2012), for four participant types: Client, DII, FII and Pro (proprietary). Long share = long ÷ (long + short) index futures contracts. Option figures are contracts long minus contracts short, calls and puts separately. Contract counts aren't adjusted for lot-size changes.

Study. Days are split into five equal groups by that day's net flow, and NIFTY's change that day is compared across groups. For the 20-day view, the 10% of days with the lowest and highest 20-session net flow are compared with all days on NIFTY's return over the next 5 and 20 sessions. Correlations are Pearson coefficients over the whole period. Windows overlap, so neighbouring days are not independent and the day counts overstate the evidence. Rupee amounts aren't adjusted for inflation or market size, so recent years dominate the largest groups. The figures describe what happened; they don't predict it.