It’s true! The FT – and social media – really do move markets; Wikipedia visits and Google search linked to swings, research says

January 3, 2014 5:26 pm

It’s true! The FT – and social media – really do move markets

By John Authers

Wikipedia visits and Google search linked to swings, research says

Newspapers report the news. They should never aim to be part of the story themselves. But in the stock market, that division is hard to sustain. A rigorous statistical study by a group of academics at Warwick Businss School has now shown that we at the Financial Times regularly move the markets we write about. (A similar exercise for our best-known competitors would surely yield the same result; it happens that this study covered the FT.)The research is part of a growing effort to understand how to interpret people’s use of data, and the trails they leave on the internet through search engines and social media, to predict how they, and markets, will behave.

The study looked at 1,821 FT issues published in the six years from 2007 to 2012 – years that included a historic stock collapse and a subsequent dramatic rebound.

The researchers counted all mentions of the 31 stocks that were part of the Dow Jones Industrial Average during this period and linked their mentions in the paper to their share price performance the next day. It showed a strong correlation – a mention in the morning’s FT meant a greater volume of trading.

Interestingly, their study involves the printed edition when much of the news on which the FT reported would already have appeared online the previous day. So the news continued to have an impact on the next day. As there are now many sources that should move markets more swiftly than a printed newspaper, this also implies that it was the news itself, rather than any editorial choice about publishing stories, that moved prices.

Financial stocks, led by Bank of America, were most likely to be news-driven during a period when markets were driven by the financial crisis.

The same researchers have been following investors’ behaviour on the internet, particularly on Google and Wikipedia, to gauge how people go about making decisions and whether this predicts market behaviour.

The findings were clear-cut. Increased visits to the Wikipedia page about a company were associated with subsequent falls in price. And search activity on Google was powerful as a gauge of sentiment.

One example is already famous. They monitored Google searches involving the word “debt” from 2004 to 2011 and built a crude market-timing model from it. If searches for “debt” increased one week, they would sell the Dow Jones Industrial Average short the next week, profiting from a fall. When “debt” searches fell, they bought the Dow the next week.

 

The results are shown in this chart. This strategy produced a gain of 326 per cent. Just holding the Dow over this period would have returned 16 per cent. And using the same strategy with the word “culture” instead of “debt” returned 5 per cent.

This study was made with hindsight. Anyone using social media to time markets will need a good idea of what themes could matter most. But results for words that frequently appeared in the FT, and thus had obvious financial relevance, were stronger.

So working out how to use social media is hard but the conclusion of Tobias Preis, who led the research, is robust: “Increasing information-gathering in financially relevant information on Google and Wikipedia is linked with subsequent stock market losses.”

How does the news about mentions in the FT, an externally driven event, affect these findings? It could be in one of two ways. As many stories are about share price moves that have just happened, the fact that the effect continues on the next day bears out that stocks tend to have momentum once they have started moving.

But trading in reaction to news could also bear out the notion of market efficiency – that prices adjust to incorporate all known information, and follow a “random walk” in response to new news. The bubbles of the last two decades have shown that there is a lot wrong with the “random walk” as a theory – but it is reassuring to confirm that traders do respond to news from the real world.

This is particularly true now, as opinions are polarised and distorted by long-running attempts by central banks to support asset prices. Bulls, who have been optimistic, now project that stock markets will be flat or down this year. That is because they think the US economic recovery is real and that the Federal Reserve will have to remove stimulus quicker than the market now expects – that could push share prices down.

Meanwhile long-term “bears” who believe the rally since 2009 is mostly a response to the Fed, think stocks could rise much more, until they form a bubble. This is because they think the Fed will have to continue to inject cheap money.

The outcome will depend on myriad decisions by investors around the world. But it is comforting that social media now give us better ways to spot changing sentiment in advance and that stock markets are still driven by news from the real world. In 2014, we should watch how sentiment is moving – and also read the news.