Quant analytics: Is anybody using social networking data for sentiment analysis in their trading strategies?
Countless people use sentiment analysis with news articles to detect market mood, it seems now many people are trying to integrate publicly available social data (from Twitter for example) into trading strategies; using sentiment as a trade indicator. Is anybody here doing this? Do we think there is any significant advantage here or is it likely to consist of a lot of ‘junk’ posts that just add to market noise and make life harder? Would love to hear your thoughts on this in terms of ways to successfully implement this kind of analysis into a strategy.
We are blending in Social Media data into the TSL Learning Engine which will export Strategy code for HFT, MFT and LFT. The engine will decide how or if to use the data in the autodesigned mathematical structure during design time.
we have quite a bit of experience at the coal face of this if its helpful
in our work for clients since 2007 we’ve been correlating financial metrics (shareprice, NPS and revenue) to social KPI’s we measure using a proprietary methodology we developed with the LSE (specifically share of voice, aggregated brand sentiment and sentiment by topic)
most of us are ex quants, we have over 5 years of data and we’ve just had M&C Saatchi buy a minority stake in us
what our data suggest – is share of voice and revenue correlate more closely at times when brands do, specific brand awareness / repositioning work – but those correlations don’t hold
its also true to say, that brands with an emotional component to their architecture – as in – tribal brands – have stronger correlations between sentiment and shareprice which again, makes sense
to say there are direct links between real time data from Twitter and shareprice…..is, in our experience – a bit of a leap to say the least – if you dig into the Derwent capital fund you find a pretty vanilla HFT algo. with a bit of twitter for PR effect (is my reading)
fascinating area of research though – we’re just about to deliver an alpha version of a cloud based solution, to find stronger correlations in our data and shareprice (based on an actuary model) – if that data is solid – we intend to start trading a small account
anyone who would be interested in a genuine collaboration – i.e. contributing stats, HFT expertise and sentiment analysis thinking (the head of the qual. dept. @ the LSE sits on our board – so we know a thing or two about discourse analysis and thematic breakdown – but we’re always keen to share and learn) – we’d be interested to hear from
NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!