New research shows that Google search results can predict movements in the market. Jeff Benjamin discusses.
When a team of academics last week unveiled research showing that Internet search traffic could be used as a leading indicator of the stock market's direction, it initially sounded a little far-fetched for an actual investment strategy.
Then, in the same week, a tweet was sent from a hacked Associated Press Twitter account reporting that a bomb had gone off at the White House, and the Dow Jones Industrial Average dropped 145 points within minutes.
When the tweet was promptly verified as false, the Dow recovered as quickly as it fell, but the message was clear: The false AP tweet at once endorsed the imposing influence of Internet activity on the financial markets and blew right past the methodology of the academic research report, “Quantifying Trading Be- havior in Financial Markets.”
The study, which expanded some earlier research on the relationship between individual company searches and stock performance, focused on Internet searches for financial terms such as “portfolio,” “inflation” and “economics.”
WISE TO THE WORD
For example, the formula was back-tested on the Dow for the period from January 2004 through February 2011 and found that going long or short the benchmark based on Internet searches for the word “debt” would have generated a 326% cumulative return. By comparison, just holding the Dow over that same period would have produced a 16% cumulative return.
“Online resources can be used to say something about investor behavior,” said study co-author Tobias Preis, associate professor of behavioral science and finance at the Warwick Business School in the United Kingdom. His co-authors are Helen Susannah Moat, a senior research fellow at University College London, and H. Eugene Stanley, a professor of physics at Boston University.
The research employed a relatively simple investment strategy of selling short the index whenever searches for the word “debt” increased during the previous week. If searches decreased, regardless of the actual number of searches, the index would go long for the upcoming week.
Among the key lessons, according to Mr. Preis, is that investors tend to do a lot more due diligence before selling an investment than they do before buying.
“It is a general feature of human behavior that it is more complicated to sell a stock that you already own,” he added.
There is a certain logic behind the idea of doing more research before selling than before buying an investment, said Beverly Flaxington, principal at The Collaborative, which provides business advice to financial planning firms.
“It is a general human tendency that to sell will take a lot more competence, because the mindset is that once we own it, we own it,” she said.
HARD TO TAKE A LOSS
The emotional challenge of selling — especially at a loss — is significant even for professional money managers, said Kent Croft, chief investment officer at Croft Funds Corp.
“As a professional, you try to make more-rational and non-emotional decisions that are based on the numbers, but we know that people will sometimes postpone the decision to sell even if it can give them a loss that they can take advantage of for tax purposes,” he said.
Can this kind of predictable human behavior be applied to an actual investment strategy?
Mr. Preis is hopeful, which is why he has formed Artemis Capital Asset Management GmbH. There are no specific details on how he plans to leverage the search engine data.
OLD-FASHIONED ALREADY
This brings us to last week's fake tweet from the hacked AP account.
We now know that the sudden market spike and rebound were attributed to computer-generated trading activity programmed to react to specific words and phrases flashing across the cyberuniverse.
With the Securities and Exchange Commission recently ruling that companies can turn more frequently to the Internet and social media for public reporting activity, the algorithmic trading programs should be licking their digital chops.
Of course, this doesn't necessarily mean the likes of Mr. Preis and his attempts to forecast market directions based on human instincts gleaned from online search activity is heading for extinction. It just means that it is already looking like an old-fashioned way of doing things.
