8.6.3 What is the informational content of the event?
Once we see an event, how do we systematically decide whether it is significant? The
number of news items for each company are too high to regard every news report as an
event. One approach could be to wait for analyst forecast changes following the news,
though, as we have shown, analysts may be slow to react or reluctant to act at all.
Alternatively, we can define significant news by its short-term price reaction, though the
secondary issue is deciding which thresholds to use to define positive and negative news
so that we do not react to every piece of information.
8.6.4 What is the holding period?
If we buy stocks following positive news flow and then discover two days later that there
is some mildly negative news, should we close the position after the two days? Such an
approach would impose excessive turnover and limit the breadth of the strategy as there
would only be a few stocks in the portfolio at any point in time.
Here we highlight a simple event-driven strategy that takes advantage of earnings
revisions momentum. Historically, earnings revisions have worked as a factor because
earnings revisions and earnings surprises tend to be followed by more revisions and
surprises in the same direction. In other words, stock prices tend to underreact initially
to the information in these events, and then drift in the direction of the event over time
(see Xu, 2008).
Here we show how investors can gain an advantage by using news flow to trade ahead
of analyst revisions. We begin by defining an information event as either good or bad
news based on the 1-day sector-relative price reaction around the news citation. We then
estimate the threshold breakpoints using the returns over the past three months and
group the subsequent period event returns into three baskets to define positive, neutral,
and negative news. Since the thresholds for the baskets change on a rolling basis, we take
into account the changing relative importance of different types of news over the
business cycle. We initially focus on all news events and then restrict our analysis to
accounting-related ‘‘hard’’ vs. strategic ‘‘soft’’ news.
Our strategy is to buy companies with positive returns around news events, sell those
with negative returns and ignore companies which are considered to have neutral news
flow. We assume that stocks are bought at the closing price on the day after the news
release and hold them for 20 days, so as to increase the breadth of the strategy. If we see
another news item within the 20-day window, we only close the position if the stock
switches from the top/bottom basket.
Our results show that investors can exploit news flow datasets by either trading
directly on news flow or combining the dataset with earnings momentum factors. Table
8.2 shows that such a strategy has generated an annualized long/short return of 14.5%
pa with an information ratio of 1.85 (pre-costs). Investors who can react quickly can
benefit from the short-term momentum following news (in particular, accounting-
related news). Given the daily nature of this strategy and consequently high turnover,
a more practical approach may be to apply news flow as a filter on existing investment
processes (using stock-specific news flow as a technical overlay to identify overbought/
sold companies).
The impact of news flow on asset returns: An empirical study 225