How sharing patterns relate to stock price

After analysing attention metrics for publicly traded companies we began looking for meaningful correlations with other performance data. The big obvious question to ask was this:

Do sharing patterns of news about companies have a relationship to changes in stock price?

We were very surprised at how quickly we found a positive result:

Sharing of coverage about companies appears to have an inversely proportionate relationship to changes in stock price. Source: Kaleida Data, March 2017

For the first 3 months of 2017 the FTSE 100 have moved in an inverse pattern to sharing volume. It’s unclear if media coverage affects stock price, but it appears that sharing of media coverage does.

The pattern in the pattern

What we’re finding with Kaleida is the act of sharing may have more substance to it than people have understood in the past. There have been indications of this with Brexit and Trump.

When people amplify news they are affecting the subjects of those stories.

Sharing patterns for coverage of publicly traded companies have an ebb and flow to them like anything else. Sometimes attention across the media is led by the subject saying or doing something. Sometimes attention is led by journalists who uncover something. And, crucially, sometimes attention is led by people who find a story interesting.

It’s in this latter class of behaviours that something triggers changes in the other direction. Instead of activity originating from the source that people respond to, there’s activity originating from people that affects the source.

Publishers all care very much about the impact of their journalism, but finding a clear line from a story to an outcome is much harder than it might seem.

Perhaps the problem is the media-centric perspective of that question. Perhaps the answer is more about what impact people are having on the subjects of the journalism they consume.


About the data

Kaleida analyses coverage from leading publishers in the US and UK. We’re looking at how they treat their stories, clustering articles from different publishers, connecting coverage to other data sources, tracking how coverage gets shared on Facebook over time, etc.

So far, we’ve collected and analysed half a million articles from about 20 news organisations, built profiles for about 100,000 subjects and tracked nearly 1 billion Facebook shares.

In our initial research on this idea we’ve only seen the correlation on aggregate data. In the same way that we’ve found that sentiment analysis works in aggregate much better than it does on a single story, we see the pattern on groups of companies but not on any single company’s stock price.

Sainsbury’s, for example, doesn’t show the pattern on its own. The FTSE 100 index, however, fluctuates inversely proportionate to the sharing patterns of all articles about FTSE 100 companies.

It’s also true that networks of things appear to rise and fall together.

In this case, where we’re looking at changes in stock price, the FTSE 100 overall seems to be affected by the sectors that get the most media coverage — supermarkets, retail, fashion, and travel. Companies in those sectors are in the news more, and the stories about them are shared more on Facebook.

This collective rise and fall is similar to the way Trump affects his daughter’s fashion businesses with the attention he draws.