The Wiley Finance Series : Handbook of News Analytics in Finance

tick time periods). We note that the relationship between squared 5-second returns and
the realized volatility over the period½t 1 ;t 2 Šis as follows:

vt 1 ;t 2 ; 5 ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1

t 2 t 1

5

X

t 1 t 2

ðr; 5 rrÞ^2

v

u

u

t 

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1

t 2 t 1

5

X

t 1 t 2

r^2 ; 5

v

u

u

t ð^3 :^3 Þ

whererris the average return in period½t 1 ;t 2 Š.
As with NewsScope alerts, volatility exhibits strong seasonalities on intradaily and
intraweekly timescales (see Figure 3.1). As one might expect, these seasonalities are only
found in the squared returns, not in the returns themselves. This discovery raised the
specter of specious results based on the correlation between news alert seasonality and
FX volatility seasonality (as measured by squared returns). This potential difficulty is
dealt with in Section 3.5, where seasonality is removed. For additional analysis of the
properties of the Thomson Reuters FX dataset see Section 3.A.1 on p. 100.

3.4 A framework for real-time news analytics

The core of our real-time news analysis engine relies on a scoring method that assesses
the relative volume/significance of news from a specific category of news. For instance,
we wish to identify periods when the volume of news about foreign exchange markets is
abnormally high, or when there is a flurry of macroeconomic news announcements.

Managing real-time risks and returns: The Thomson Reuters NewsScope Event Indices 77

Figure 3.1.Average realized volatility of the CAD/USD exchange rate return over the course of a
week, averaged over 5-minute (gray) and 30-minute (black) timescales. Note the strong daily peaks.