Topics: social media

MIT Technology Review: “This week’s fleeting stock market crash prompted by a false report from the Associated Press’s hacked Twitter account has focused attention again on the growing Wall Street practice of mining news and social data to make trades.

A study in Nature Scientific Reports today illustrates just how lucrative the right combination of algorithms could potentially be.
Using Google Trends, researchers analyzed the Google search query volumes from 2004 to 2011 for a set of 98 mostly finance-related search terms, looked at how stock prices changed over that same time, and tried to see if they could retroactively tease out search patterns that showed “early warning signs” of market moves. They also tested trading strategies that would act on these signs.

The volume of the search term “debt” turned out to be the word that showed the most promise, and one trading plan based on changes in searches for this term would have yielded a return of 326 percent over the period analyzed, the authors found. For comparison, a “buy and hold” investment in the Dow Jones Industrial Average yielded 16 percent return.”

Nick Bilton in NYT: “It looks as if the theory of the “wisdom of crowds” doesn’t apply to terrorist manhunts. Last week after the Boston Marathon bombings, the Internet quickly offered to help find the people responsible. In a scene metaphorically reminiscent of a movie in which vigilantes swarm the streets with pitchforks and lanterns, people took to Reddit, the popular community and social news Web site, and started scouring images posted online from the bombings.
One Reddit forum told users to search for ”people carrying black bags,” and noted that “if they look suspicious, then post them. Then people will try and follow their movements using all the images.” In the process, each time a scrap of information was discovered — the color of a hat, the type of straps on a backpack, the weighted droop of a bag — it was passed out on Twitter like “Wanted” posters tacked to lampposts. It didn’t matter whether it was right, wrong or even completely made up (some images posted to forums had been manipulated) — off it went, fiction and fact indistinguishable. Some misinformation online landed on the front page of The New York Post, incorrectly identifying an innocent high school student as a suspect. Later in the week, the Web wrongly identified one of the suspects as  a student from Brown University who went missing earlier this month…
Perhaps the scariest aspect of these crowd-like investigations is that when information is incorrect, no one is held responsible.
As my colleague David Carr noted in his column this week, “even good reporters with good sources can end up with stories that go bad.” But the difference between CNN, The Associated Press or The New York Post getting it wrong, is that those names are held accountable when they publish incorrect news. No one is going to remember, or punish, the users on Reddit or Twitter who incorrectly identify random high school runners and missing college students as terrorists.”