WiseWindow provides a service used to predict market trends--in this case stock market trends--based on customer sentiment expressed on social media sites, blogs, and message boards, rather than its more common applications to market research. The initial data feed available through Bloomberg focuses on the airline industry, where WiseWindow's consumer sentiment data has proven to have a high predictive value, the company said. WiseWindow will introduce additional industry-specific data feeds in the coming weeks.
According to an analysis performed by advanced analytics consulting firm Emerald Logic, factoring in WiseWindow data on top of a simple momentum trading program boosted returns by over 30% on an annualized basis for GM, Ford, and Southwest Airlines. For American Airlines, the uplift was 65% annualized. Volatility of returns was also significantly reduced.
"We're not claiming this is the only variable--that would be foolish," WiseWindow CEO Sid Mohasseb said. Wall Street firms and hedge funds would typically add the data as one additional input to their own proprietary predictive models. The data would also be weighted most heavily for stocks where consumer sentiment makes the biggest difference, as opposed to industrial stocks whose brands are seldom mentioned on Twitter or elsewhere in social media.
"The farther you get from the consumer, the more questionable the data becomes," Mohasseb said. Yet within the consumer realm, sentiment analysis does a good job of predicting both stock prices and sales trends, he said. The stock market correlation is "upwards of 0.88" two to three weeks in advance, while the correlation to sales is in the range of 0.7 to 0.8 for a two-week forecast, he said. Correlations are measured on a scale from -1.0 to 1.0.
Emerald Logic CEO Patrick Lilley, whose firm does much of its work for the financial industry, said he has no formal partnership with WiseWindow but was intrigued by the results it claimed for predicting stock market trends and agreed to provide independent verification. He took a simple momentum trading model that by itself was capable of beating the market and then used the WiseWindow data as a sort of filter on the results.
"If the model said buy, I would check whether I really should buy based on the WiseWindow data," Lilley said. If consumer sentiment was running against the company whose stock was recommended, he would eliminate it from his list of stock buys. "I found that it eliminated more bad trades than it eliminated good trades," he said, and that's where the statistics for improved stock picking results come from.
Lilley related this approach to the advice from a recent BrainYard interview with Deloitte analytics expert John Lucker, in which Lucker suggested that social media analytics need to be checked for accuracy against secondary and tertiary sources.
"The stock market can be very unforgiving--if you get it wrong, you get it wrong--but WiseWindow did very well against that tough benchmark," he said.
Attend Enterprise 2.0 Santa Clara, Nov. 14-17, 2011, and learn how to drive business value with collaboration, with an emphasis on how real customers are using social software to enable more productive workforces and to be more responsive and engaged with customers and business partners. Register today and save 30% off conference passes, or get a free expo pass with priority code CPHCES02. Find out more and register.