Varance Swaps as predictor of future volatility?

In theory, VS strike represents the market prediction of realised variance over the term of the swap.

- complicated by supply/demand factors --> especially at longer maturities where driven by structured product flow. 

- at shorter maturities, risk-aversion can bias the VS strike to be > "fair" expectation of future realised variance. 


At least at short dates, implied variance tends to do a relatively good job of forecasting future realised variance.

Back-testing --> implied variance better predicts changes in future RV than previous RV. 


Variance --> biased estimator --> generally overestimates realised volatility. 

Given historical data, a simple model of future realised volatility as 0.9* VS strike provides the best fit with the data. 


What extra information could the variance swap and option markets be pricing in?

  • Mean reversion of volatility: the implied variance can take account of the current medium term “average” of volatility and the fact that volatility tends to be mean-reverting within a regime.

  • Knowledge of forthcoming events: volatilities (especially short-dated) will tend to be marked up ahead of earnings seasons or other events likely to increase volatility. In this sense implied variance can be truly forward looking, whereas past realised volatility is not.

  • Risk aversion in the market: If investors are nervous and willing to pay extra for protection, they are perhaps also prepared to liquidate their positions quickly in a downturn and increase market volatility. Thus an increase in implied variance could precede an increase in realised volatility if the market sells off. 


Further in the future is harder to predict.

Any edge of implied over realised volatility as a predictor for future realised  disappears for maturities longer than 3m. 


Correl iv past future rv

Add a comment