Let me fix the title: <i>Hedge fund claims to use Twitter to predict stock market in an effort to attract attention from investors.</i><p>Other variations on this theme that I've encountered:<p><i>Hedge fund uses all-Apple architecture to gain competitive advantage.</i> (Translation: a platform-independent runtime [1] on top of XServe + imac/mac pros for workstations. )<p><i>Hedge fund uses NoSQL solutions to predict the market, minimize risk.</i> (Translation: flat files for quote-level data, same as everyone else [2].)<p>I'm sure they throw twitter sentiment into their machine learning program as one piece of data. The learning system might even give it a non-constant coefficient. But I'd be really surprised if it plays a major role.<p>[1] I won't name the system, but it's something along the lines of (and might even be one of) the JVM, CLR or Smalltalk image.<p>[2] I'm being unkind - some banks use ADABAS and other such systems from the 70's.