Trading shares during the day that a company releases its earnings reports by going long on ones with positive news and short on the ones with negative news can generate a daily return of 1.3 percent.
That is the bold conclusion of a study conducted jointly by the Amundi Institute and the Toulouse School of Economics. Using artificial intelligence software from Causality Link and a long-short portfolio, the researchers were able to confirm that news signals influence short-term price movements.
“A promising area of development in the asset management industry,” the researchers said. “Investors can use these news signals to develop strategies aimed at achieving higher gross returns.”
News signals
Using an algorithm, the researchers mined textual data available in financial markets. Articles, transcripts of calls, research from brokers: everything was analysed by using artificial intelligence, searching company names, its KPIs, the change in direction of these KPIs and its timestamp. Was the change in the past, is it in the present, or is it expected in the future?
Based on this data, the system arrived at an aggregated news signal that captured positive or negative tones in the news flow as well as measuring the popularity of the news flow around a particular company. The institute and the university then examined the returns of two portfolios on the day of publication. One consisted of stocks with the most positive news signals - the long portfolio. The other - the short portfolio - comprised the stocks with the most negative news signals. They also made subdivisions into high versus low news coverage, “fresh” versus old news, types of news (financial or ESG), the timing of these news releases, news horizon and company size.
Future matters most
In addition to the aforementioned conclusion that this type of corporate news has a direct impact on share price performance, the researchers concluded that news about the future caused much greater share price reactions than news about the present or the past.
Moreover, stocks appeared to react more when there is a lot reported about them, on news as current as possible and on purely financial news. Finally, company size matters: shares that are not part of the Russell 1000 index react more strongly to news than those that are.
Regarding the average daily return of -0.54 percentage points for the short portfolio on the day of publication of the corporate news and 0.73 percentage points for the long portfolio, the researchers add that the average returns were also positive and significant on one day prior to the day of publication.
The researchers explain this on the basis that some investors have access to private information and act on it. “This pattern may also be due to the delay in publication of some news sources, for example between transcripts and brokerage reports on the one hand and news reports on the other.”
Time is an important factor, as news is incorporated into share prices very soon after publication, the research shows.
Trading costs matter
Based on the theory from the study, an investor should therefore exit companies with positive news after a few days, take profits and move into ánother companies with positive news. Asked whether that would leave enough for returns, as it would incur a lot of costs, head of investor intelligence and academic partnerships Marie Brière replied that transaction costs are not estimated. “They are difficult to estimate, especially on the short side of the portfolio, but also because in practice they depend on the size of the portfolio.”
In the paper, the researchers assume a portfolio of 15 to 123 stocks, with an average of 70. And of a high daily turnover rate. “In most of our subsamples, the turnover rate is greater than 80 percent,” the researchers write in the paper. A turnover rate of 100 percent would mean that portfolios formed on “day t-1” - the day before the stock price can develop on, say, aftermarket news - would have to be completely liquidated, to form a new one on “day t”.