The Bank of England built a Twitter tool to predict bank runs but it screwed up for a really weird reason

The Bank of England has revealed on its blog that it built a tool in the run-up to the Scottish independence referendum last year to try and predict a run on the banks using Twitter.

The central bank was worried that a "yes" vote could provoke a rush of people pulling their money out of Scotland and thought a good way to monitor the situation would be to look at tweets.

Here's the bank:

A key differentiator from traditional media is timeliness: any individual can tweet from almost any location at any time. And those tweets can be captured and analysed almost immediately, too. The flow of data continues 24/7, allowing you to monitor sentiment after markets close and before they reopen.

Working with social media analysis company DataSift, the Bank built a tool in a few days that could collect and analyse tweets in realtime, filtering for certain key phrases that they thought could signal a bank run.

But one weird anomaly completely thew the system out.

Here's the data the tool collected in the run-up to the vote, charted:

You can clearly see a big spike in the early hours of September 15. The Bank was initially worried but realised the spike had nothing to do with the Scottish Referendum - it was caused by American Football.

Here's they are again:

On closer examination, it transpired that we were looking at a series of tweets and retweets involving the Minnesota Vikings. This had been captured because they combined the term "run" and the abbreviation "RBs". But in this context, the reference was to Running Backs and not the Royal Bank of Scotland! To avoid this pitfall, the search terms were subtly changed to avoid that particular pattern of upper and lower case characters.

In the end, the Twitter monitoring didn't come to much but the Bank says it was "a valuable exercise, building capabilities and knowledge to serve as a foundation for future projects."

While Twitter can be a difficult mountain of data to scale, it does contain valuable financial information. The European Central Bank said in a recent study that it analysis of Twitter can give "statistically and economically significant predictive value" for stock markets.