WallStreetBirds taps Twitter data for stock tips

Modulus Informatics has launched a free online service providing stock trading analysis based on Twitter sentiment data.

Research published in October found that analysing the content of daily Twitter feeds using two mood tracking tools enabled the team to predict with an 87.6% accuracy the daily ups and downs in the closing value of the Dow Jones Industrial Average.

This prompted London-based Derwent Capital Markets to launch a £25 million social media-based hedge fund, using Twitter to gauge market sentiment.

Another academic study from Technical University of Munich economists also found that the sentiment of tweets is associated with abnormal stock returns. The academics are now using Twitter for a Web site that predicts stock price trends.

Modulus Informatics has now joined the fray with its free site WallStreetBirds.com, that lets users pick stocks to analyse before its system pulls in Twitter data and comes up with buy or sell signals.

The firm is also providing a programming API that it says can be used for developing high frequency trading systems based on the instantaneous analysis of social media data.

Finextra Verdict: We hear much about Twitter's fabled prescience in forecasting news and trends, but on some days the world beyond the shallow tweet chatter seems very far away. Stock picking engines rely on their ability to sift throught the general noise and zero in on the hard signals that matter. But what if Twitter has become the preserve of the chattering classes, clogged with an endless stream of self-congratulatory tweets from total strangers pretending to be best mates? Evidence for the prosecution: As the debt crisis rumbles on and people run riot in London, what were the top trending topics on Twitter this morning?

We rest our case, and don the tin hats in readiness for the barrage (we will be tweeting this).

Channels

Wholesale banking

Keywords

Dealing rooms Social media

Comments: (2)

Elizabeth Lumley - Girl, Disrupted - Crayford 08 August, 2011, 15:59

Be the first to give this comment the thumbs up

0 likes

Really? Because I have #LondonRiots trending on my UK trending chart and over the weekend it was #woodgreen.

"Stock picking engines rely on their ability to sift throught the general noise and zero in on the hard signals that matter." That is exactly it. That is how these 'stock picking' engines work.

The argument "But what if Twitter has become the preserve of the chattering classes, clogged with an endless stream of self-congratulatory tweets from total strangers pretending to be best mates?" doesn't make sense when talking about a 'stock picking' service. It's kind of like saying: 'that new fruit & veg stall down the road will never work, because that poncy pub on the corner is full of wankers.' The reasoning doesn't link up.

I guess two academic studies should be disregarded? The real proof is if any of these firms/services make money. I just hope if that happens no one celebrates with a 'self-congratulatory tweet'.

Report abuse

Steve Ellis - Finextra Research - London 10 August, 2011, 08:50

0 likes

Twitter is just one place that conversations take place. There are lots of other venues too (both online and offline). But Twitter happens to be public and available in a easily analysed format. I wouldn't write it off so quickly.

I'm guessing there are lots of people looking at this potential use of social data. Using Twitter (and other sources of public, conversational data) for stock trading is an issue of being confident in sentiment analysis, and the ability to differentiate between noise and signal.

Brands have been trying to apply similar technologies to determine issues of consumer perception, sentiment and real time reputation management for quite a few years now. Most are doing something with one of the multiplicity of vendors in the area. But no-one has really nailed it yet. They tend to sound good on paper but scratch at the surface of sentiment analysis, and the data doesn't really stand up in court, or it requires considerable human interpretation to be added.

I'd suspect a similar evolutionary cycle with the application of stock trading using social data. The great thing about stock trading though - as Liz points out - is that there is a definitive measure of success: did it make money? Whereas insight on sentiment will always be subject to debate and contrary viewpoints.