The Wiley Finance Series : Handbook of News Analytics in Finance

(Chris Devlin) #1

In light of the contrasting performance of the event studies for realized volatility and
implied volatility, we study the differences between these quantities that would explain
this behavior. First, we note that implied volatility is imputed from options prices by
inverting the Black–Scholes (or other similar) options-pricing formula. Thus, any fore-
cast power with respect to implied volatility would imply forecast power for options
prices, and since these options are actively traded we would expect only very modest
price inefficiencies with respect to news.
Second, we observe that implied volatility is meant to reflect the expected volatility
over a lengthy time horizon (e.g., 1 month), whereas the analysis of Section 3.5 concerns
volatility over a much shorter period (e.g., 1 hour). It has been documented that implied
volatilities tend to be much better predictors of long-term future price volatility in equity
markets, and we find the same effect in foreign exchange data. Indeed, Figure 3.8 plots
the correlation between future realized volatility (for horizons from 1 minute to 10 days)
and the current implied volatility. For comparison, the grey curve in Figure 3.8 plots the
correlation between (past)t-hour EUR realized volatility and the futuret-hour realized
volatility. We note that the correlation between implied volatility and realized volatility
decays as the time horizon gets shorter, and once the time horizon is less than 2 hours,
historic volatility outperforms implied volatility as a predictor of future realized vola-
tility. Implied volatilities from other banks, as well as 1-week implied volatilities, showed
similar behavior. These findings confirm those of Pong et al. (2004) and Taylor (2005).


Managing real-time risks and returns: The Thomson Reuters NewsScope Event Indices 91

Figure 3.8.Correlation of 1-month Socie ́te ́-Ge ́ne ́rale-quoted EUR implied volatility with future
t-hour realized volatility (black line). Correlation of (historic)t-hour EUR realized volatility with
futuret-hour realized volatility (grey line). All realized volatilities are de-seasonalized (as described
in Section 3.5).

Free download pdf