Greece? Zzzz. EU bank says TWEETING can move the stock market

It must be true, Big Data says so

By reading social media blather we can predict the stock market? Hmmm

What does all this mean? Something might be very useful in practice but does it concord with theory? Here the paper assumes it's an indication that behavioural finance is a better description of market pricing than the efficient markets hypothesis is. Recall that this is not the statement that markets are efficient, nor that we should do everything using markets. Rather, it says that markets are efficient at processing the information about what prices should be.

If people start sending twats around stating that the market might go up, then other people are convinced by this into buying – which puts the market up.

I sorta have my doubts about that because they're only talking about a few hundred instances of “bullish” in a twat leading to a rise. Can't see that minor a persuasion leading to significant buying.

However, it would be entirely consistent with a not-very-efficient reading of the efficient markets hypothesis, which is that markets are efficient, yes, but not instantaneous. There does need to be some time period in which the information about what prices should be is processed in. At which point we could say that the Twitter optimism was less a cause of the market rising and more a symptom of that information being processed that the market might or will rise tomorrow.

The end point of their paper, and the real point about this big data stuff, though, is that we can indeed spot correlations by crunching through that information, but deciding causality is rather more difficult:

While our study shows a promising predictive correlation between Twitter bullishness and stock market prices, it offers no information with regard to causality. Causal inference is important for result interpretation, robust prediction and policy-making. Drawing causal inferences from Big Data is a challenging research problem. Consequently, future work will focus on developing a novel theoretical framework by combining experimental design methods and machine learning algorithms to infer the causal relationship between Twitter bullishness and financial markets.

Well, that's great. Apparently the European Central Bank has something interesting to do rather than trying to solve Greece's problems. ®