Implied correlation: what the index IV and 50 stock IVs say about how stocks move together.

• Implied correlation0.39on 10 Jun 2026, 30 Jun 2026 expiry
• Realised, 20 sessions before0.18to 10 Jun 2026, known that day
• Realised, 60 sessions before0.40to 10 Jun 2026, known that day
• Realised over the option's life0.2813 sessions, 10 Jun 2026 to 30 Jun 2026

On 10 Jun 2026, the 30 Jun 2026 Nifty options were priced at 15.03 vol points. The 50 stocks in the set, weighted by the fund holdings, average 23.3 vol points, and would add up to 4.7 if no two of them moved together. An index priced at 15.03 sits at a common correlation of 0.39 between those two ends. The same formula applied to what the stocks and the index then realised gives 0.28 over the 13 sessions to expiry, against 0.18 over the 20 sessions before and 0.40 over the 60 before.

Volatilities are in vol points a year. Implied rows use at the money implied volatilities of the 30 Jun 2026 expiry; realised rows use close to close volatilities of the Nifty 50 and of each stock, 252 sessions a year. The 20 and 60 session rows were known on the day; the last row was not, and is the one the implied figure was a price for. A stock whose price moved by more than 25 percent in one day inside a window (a corporate action or a crash) is left out of that window and the weights renormalised: here TRENT. 13 sessions is a short sample, so the realised figure over the life moves a lot with one or two days.

The largest names

Weight is the share of the set after renormalising. ATM IV is the at the money implied volatility of the stock's 30 Jun 2026 option, in vol points, plus your shift. Realised columns are annualised close to close volatilities in vol points.

How much the number moves

Rows move the index IV, columns move every stock IV, both in vol points.

One vol point on the index is worth about 0.06 of correlation here, and two vol points on every stock about 0.07 the other way. A small difference in the IV of one index expiry therefore moves the correlation by a lot: read it to one decimal, not two.

Method: with weights w (summing to 1) and stock vols s, the index variance is the sum of (w s) squared plus the correlation times the cross terms, so ρ = (σ² − Σ(w s)²) / ((Σ w s)² − Σ(w s)²). The weights are a proxy: the holdings of a Nifty 50 index fund at the month end before 10 Jun 2026, times each stock's NSE closing price on the day, not the index provider's own file. Each stock is represented by its at the money IV alone, so its smile, its dividends and the difference between the index options' skew and the stocks' skews are left out; the index IV is the at the money IV of the same expiry. One common correlation for all pairs is a simplification: the real matrix has high and low pairs, and the number is an average in that sense. Weights, IVs and realised volatilities come from end of day files, not from a quote you could trade at. A Nifty Bank version is not offered: the data holds no weights file for its twelve stocks, and a stand in built from the six banks that are also in the Nifty 50 would not be that index. This page cannot trade.