The idea that sentiment indicators based on Tweets can accurately predict the market has been galvanizing, to say the least. Yet despite some academic evidence that this is the case, there are plenty of skeptics.

To that list we can add John Bates, the chief technology officer at Progress Software. He offers some interesting thoughts on why the use of Twitter-based indicators may be less logical than some social media enthusiasts want to believe. Part of his argument is that most of the "sentiment" being indexed via Twitter amount to lagging indicators. If there's big news, Tweets about the news will spike, but the provenance of those Tweets are likely to Reuters or Bloomberg. When it comes to macro events, Bates wonders if the "consciousness" of the event--if measured via Twitter--will take too long to develop, especially for traders who need to trade the event instantly.

Not everyone will agree with Bates, and they argue with evidence. Derwent Capital Market has set up a small hedge fund that will invest according to Twitter-based indicators derived from work by two academics at Indiana University. We'll see how that fares. My sense is that Twitter can be effective as one of many sentiment tools at a trader's disposal. We've suggested that it may work best as part of a machine-readable news service. Modulus Informatics' Wall Street Birds, a free service that offers sentiment indicators built on Tweets, presciently comes with an API set that allows for firms to build a Tweet component into its current models. This strikes me a sound approach.

For more:
- here's the column

Related articles:
Free service offers sentiment indicators from Twitter  
Goldman Sachs banker Tweets cause stir  
Machine readable news still making gains  
First Twitter-based algo fund launched in London