related information receive the greatest share price reaction, whether the news is good or
bad. The rank correlation based on our measure of short-term returns and standardized
unexpected earnings (measured as the change in analyst EPS next 12-month forecasts
over the same window) are close to 20% for earnings-related and guidance-related news.
8.6 Designing a trading strategy
8.6.1 Turning a dataset into a trading signal
Extracting and cleaning a news flow dataset is only the initial challenge. There are a
variety of ways to incorporate this information into a trading strategy. We briefly
highlight some of the issues worth considering when taking a news signal and designing
an event-driven trading strategy.
8.6.2 How to define the event?
Companies may announce several different news items over the course of a month,
which may either be received positively, negatively, or with a muted price reaction.
Following the release of information it may be the case that the market and/or analysts
initially overreact or underreact to the news. Analysts may wait for a series of con-
firmatory news signals over a couple of months before considering changing their
forecasts. Should we react to every news item or wait for a trend in consecutive news
items?
224 News and abnormal returns
Figure 8.9.Average change in EPS estimates following news (source: RavenPack, Factiva,
FactSet, Macquarie Quant Research).