Mining Social Media for Market Insights

The European Central Bank has been particularly busy. It is engaged in an asset-buying program. Despite a European Court of Justice ruling highlighting the conflict of interest between its bank supervisory function and role as creditor, the ECB is still part of the Troika official creditors and participated in the marathon negotiating sessions over Greece. 

The ECB is closely monitoring the situation at Greek banks.  It has not increased the ELA ceiling for the second consecutive week, following the Greek government’s acceptance of the creditors’ demands.

The European Central Bank has been particularly busy. It is engaged in an asset-buying program. Despite a European Court of Justice ruling highlighting the conflict of interest between its bank supervisory function and role as creditor, the ECB is still part of the Troika official creditors and participated in the marathon negotiating sessions over Greece. 

The ECB is closely monitoring the situation at Greek banks.  It has not increased the ELA ceiling for the second consecutive week, following the Greek government’s acceptance of the creditors’ demands.

Still its staff has time to conduct some fun research.  It released a report that many market participants will find interesting.  It looked at whether Twitter and Google can be useful in predicting the stock market.  It finds that Twitter is better than Google.  It is also better than the survey that the ECB currently uses.  The ECB staff (Huina Mao, Scott Counts and Johan Bollen) concludes that Twitter generates statistically and economically significant predictive value.  The caveats are that it is particularly useful in the US, UK, and Canada.  It is not very useful in China.  In addition, the predictive value is only for the very short-term. 

The research looks at the use of “bullish” and “bearish” words on tweets and in Google.  A one-point increase in the sentiment survey index it currently uses translates, for example, into a 2.26 bp move in the Dow Jones Industrials the next day.  In contrast, a one-point increase in its Twitter index is associated with a 12.5 bp move in the Dow. 

This research found that Twitter bullishness helps predict Google bullishness.  This suggests that Twitter information precedes Google searches.   It also finds that the bump given to stocks from the Twitter bullishness wears off, and stocks return to what the ECB staff says are fundamental values within a week. 

The ECB’s work shows a positive correlation between Google queries and Chinese stocks.  It suggests that others look at Weibo, which is a Chinese-version of Twitter.  Students and practitioners may pick up mining social media for insights into other markets, including the foreign exchange market.

ECB and Twitter is republished with permission from Marc to Market