Wake up to the Twitter effect on financial markets

FT.com

How does crowd psychology affect financial returns? That is a question economists have been pondering since traders first huddled in trading pits, dealing floors or around computer screens. And this week the issue has become doubly relevant, given how markets in gold - or Bitcoins - have gyrated.

But now data scientists are jumping into the fray as well. Two academics at the MIT media lab in Boston - Sandy Pentland and Yaniv Altshuler - have been crunching vast quantities of computer data to track what happens to investors plugged into social media, such as Twitter.

This marks a new frontier for investment research; though economists could always track market returns, the advent of social media - and big data - means researchers can now track investment returns alongside information flows, with growing precision.

And the results from this analysis are intriguing. The MIT research (which has not yet been fully published) suggests investors do not perform most effectively when they are isolated from social groups. The image of a brilliant, maverick trader sitting alone, shunning conversation to make winning individual trades is wrong. But neither do traders outperform when they are embedded too deeply in any one market group (or chat room), be that the gold bug community, Japan watchers, Bitcoin enthusiasts, or anything else.

Instead, the best returns occur when investors are plugged into diverse social groups that enable them to collide with information from multiple networks. In the social media world, as in real life, it pays to hover on the edge of cliques - but not get slavishly sucked into just one.

The tool the MIT analysts used to measure this behaviour was a trading platform known as eToro. This site was created 15 months ago to capitalise on the trend among individual investors to use social media to swap trading tips and news. But eToro does not simply enable traders to chat about markets; it also lets traders “follow” each other, by monitoring other investors’ trades, discussing them in an open forum and, crucially, by copying them too, automatically if they choose.

Thus people can build a trading strategy on the back of market chatter, or by mimicking a successful fellow trader (or “gurus”, or “tribal leaders”, as Yoni Assia, the chief executive, calls them).

Now, Luddites who hate social media - or who have never watched people “flock” in cyberspace, to use the phrase coined by Biz Stone, Twitter’s co-founder - will find the thought of following a tribal trading guru bizarre. It might also look downright foolish, given that classic investment theory suggests the best returns occur when people trade on information not freely available to the crowd.

Yet 2.7m traders have flocked to eToro, performing more than 40m trades, and other similar sites are springing up.

When Pentland and Altshuler conducted studies on those eToro investors, using different control groups, they found that this received wisdom about crowd behaviour did not hold true. “Social traders” who received information from a wide variety of social groups - and copied a range of gurus - performed 10 per cent better than “normal” traders. Those “socially aware” traders also performed about 4 per cent better than traders following one or two gurus.

The MIT researchers are doing additional analysis. But the pattern that is emerging is clear: maintaining diverse social ties - and swapping information with several different crowds - tends to raise returns. And Pentland, Altshuler and their colleagues think they can project the optimal level of diversity. They have devised a system that deliberately mixes up the information flows and tweets traders receive, to create a sense of variety, and that apparently boosts returns by 209 per cent.

So what should anyone make of this? One implication is that investors or regulators need to spend more time thinking about the way social media can affect financial markets.

A couple of weeks ago, the US Securities and Exchange Commission made one (belated) step in that direction: it said companies could now distribute information via Twitter if part of a pre-announced policy.

But there is far more that needs to be done, as social media grows in influence.

The more basic lesson we also all need to reflect on the type of information we use in our own lives (or investments). The pool of traders Pentland and Altshuler studied is a relatively small, self-selecting group. But their tendency to huddle in cosy cyber tribes is not necessarily that unusual: we all like to stick with ideas or people who reinforce our views.

But tribalism in any form tends to breed tunnel vision and echo chambers, in cyber space as much as the real world. And that has a nasty habit of creating market manias and investor losses. Just ask that bruised tribe of Bitcoin enthusiasts, or those goldbugs.

Copyright The Financial Times Limited 2013