You could not escape the news, even if you wanted to. The paper is published on Arxiv.org. It’s a typical example of what our book will be about.

On the _optimistic _side, the future of the stock market is somewhere in our collective conscious, so capturing all the data about our lives should, in theory, enable one to make ‘predictions’. It is the basis of so-called crowd-sourcing.

On the _pessimistic _side, some considerations are in place:

  • Prediction does not necessarily mean ‘understanding’, or in other words, correlation is not the same as causation.

  • Finding an algorithm that does a prediction based on a huge amount of data is cumbersome, to say the least. The team of researcher found out about this particular correction accidentally, they were not explicitly looking for it.

  • As mentioned in comments on other sites (Technology Review, ReadWriteWeb, etc.), if everyone would use this algorithm for its prediction it would become useless.

  • An accuracy of  87,6% seems high, but bare in mind that a monkey (or a randomized algorithm for that matter) should get you already to 50%.

This is not the first twitter analysis that shows predictive or analytical value. It is, to our knowledge, the first that is well-founded and statistically relevant.

Comments