Can you data mine social data to predict the future? That seems to be one key question for stock pickers analyzing Tweets to predict market investments.
Professor Bernardo Huberman argues that there’s a difference between sentiment on the web versus what’s actually believed and done in real life. He calls it “the economics of attention.”
If you were to predict the election based off the Twitter stream from the last three debates you could say Obama has the election in the bag. But there’s no analysis that can solve for the moment the voter makes that final decision. Any number of things can happen leading up to the polls.
Tweets can be misleading since most people fear expressing themselves fully online, especially in an open environment like Twitter. Investors are even more cagey about broadcasting their stock picks.
Hashtags, groupthink, and advertising can also shift discussion on the web. What’s opined isn’t actually what’s believed; social media users just want to get in on the action.
There’s some predictive analysis on the web that when aggregated can positively impact strategic decisions. In short, the masses can be strikingly accurate.
The challenge therefore is balancing the organic sentiment from the bias, a true test in guessing pyschology.