Sentiment Analysis: Media vs Social Media

sentiment analysis, social media, twitter, dataminr, trading, etrading, fintech, traders

Social media can’t revolutionise the field of sentiment analysis for trading… yet.

Relatively recently, I wrote an article on the revolutionary impact that Twitter could have on stock market trading.

The post generated a lot of interest. It got plenty of views, and sparked a couple of very active debate on LinkedIn – along with the usual combination of agreement, dispute and abuse that one generally expects from posting on open online forums.

I was made aware that, previously, I neglected to acknowledge that sentiment analysis has always been a prominent trading strategy. It was around long before Twitter reared its head, and certainly before social media had the potential to impact upon stock markets. Dow Jones has been publishing an economic sentiment indicator for many years, and investment banks have frequently incorporated sentiment analysis into their practices. Come to think of it, the entire stock market is essential sentiment-driven, so trading itself is always a practice of sentiment analysis to some degree.

So what does social media have to offer that wasn’t already there? The main sentiment in reaction to my article was that social media does not, and will not, offer a new revolutionary trading strategy. I was told that the achievements of firms like Dataminr and SMA were rare successes for a system that is, in most cases, fundamentally flawed.

And the more I thought about it, the more I was inclined to agree with them. How can the opinions of a mass of individuals, many of whom are ill-informed or regurgitating the opinions of the ill-informed, benefit a knowledgeable trader? Are you even worth your salt as a trader if you need half the world to be tweeting about a financial event before you become aware of it? If you’re jumping on that bandwagon that late, surely you’ve already missed it.

Having said that, I still can’t entirely ignore the evidence present in my last post, so here’s the conclusion that I have come to.

When Dataminr and SMA were successful, it was because they used their analysis of social media to foresee events before anyone in the conventional media could. And the sentiment analysis of Dow Jones and other previous firms has been based entirely upon what has been reported in the news. If social media analysts can predict these events first more frequently, then their particular brand of sentiment analysis will become immensely valuable. The issue is that they have only done so on infrequent occasions.

Here are just a couple of improvements that need to occur before social media can usurp the conventional media as chief sentiment analysis providers:

Start ranking opinions. One LinkedIn contributor made the point that the ‘fundamental flaw’ of social media sentiment analysis is that every single opinion is equally weighted. If opinions could be somehow ranked based on knowledge, social media sentiment analysis could be vastly more accurate.

Conquer the language barrier. The conventional media still has the edge when it comes to correctly analysing ‘foreign buzz’. Social media analysts need to focus more on translating comments, posts and tweets from languages other than English, as it would offer more insight into potential activity on foreign stock exchanges.

If you’re interested in social media’s impact on trading, then you should attend The Trading Show 2015 and hear industry expert Harper Reed talk on the subject.

the trading show, 2015, london

Comments 1

Leave a Reply

Your email address will not be published. Required fields are marked *