Using Social Media to Identify Market Inefficiencies: Evidence from Twitter and Betfair

A place to discuss anything.
Post Reply
User avatar
jamdog
Posts: 33
Joined: Wed Sep 20, 2017 10:34 pm

A pretty interesting premise and findings here:
https://editorialexpress.com/cgi-bin/co ... per_id=643
So what can social media content tell us about fundamentals? Can Tweets, to take one social media output, inform us of whether an asset is mispriced? In this paper we answer these questions by analysing Twitter posts and Betfair betting prices for a season of English Premier League soccer matches in 2013 and 2014. Crucially, at the end of the each match, fundamentals are revealed as teams either win, and the bet pays out, or they lose draw), and the bet does not pay out. We can therefore analyse whether Tweets predict asset fundamentals, rather than just short-term returns. Furthermore, we can examine whether any effect is concentrated just prior to information events (goals, red cards etc.), or just afterwards when the implications of information must be deciphered. In our study, we identify the match that a Tweet refers to, and indeed the team that the Tweet focuses on, by the hashtag used (e.g. for Manchester United Football Club, one hashtag used is #mufc). This harvest of Twitter data yields 13.8 million Tweets during 372 matches, an average of 5.2 Tweets per second, during the 2013/14 season

We find that aggregate tweeting does indeed predict fundamentals but, surprisingly, this effect is concentrated in the aftermath of events (e.g. goals, red cards). Rather than breaking news faster than the betting market, activity on social media aids in the interpretation of information. To be specific, pronounced social media activity for a given team in a given match, after a goal has been scored (or indeed conceded), implies that that team is more likely
to win than contemporaneous betting market prices suggest. We also find that the predictive power of social media activity is concentrated amongst a subset of tweeters: the Tweets of British Broadcasting Corporation (BBC) journalists, in particular, are highly predictive of a team’s fundamental probability of winning, and not just after a significant match event has occurred.

Once we account
for these two factors, we find that a simple strategy of betting when the BBC have tweeted
about a team yields average returns of 2.56% from 1,615 bets, and a strategy of betting when
aggregate tone for that team is positive yields average returns of 2.28% from 903,821 bets
.
These returns — for an investment duration of no more than 90 minutes each — compare
very favourably with average returns for all bets of -5.41%, and indicate that the marginal
information contained in social media data is substantial.
User avatar
jimibt
Posts: 3641
Joined: Mon Nov 30, 2015 6:42 pm
Location: Narnia

nice find and in truth, if you were to harvest crowdsourced data for similar events, you may find that the interpretations varied significantly due to both the more balanced nature of the masses (i.e. a potentially better view of the market) and the more immediate impact of the news..

definitely the way fwd tho..
User avatar
jamdog
Posts: 33
Joined: Wed Sep 20, 2017 10:34 pm

It would be very exciting if this was replicated over several studies. It would definitely be a saleable service if someone were to aggregate social media responses in real-time. Be interesting to see how this develops.
User avatar
jimibt
Posts: 3641
Joined: Mon Nov 30, 2015 6:42 pm
Location: Narnia

jamdog wrote:
Wed Apr 04, 2018 11:18 pm
It would be very exciting if this was replicated over several studies. It would definitely be a saleable service if someone were to aggregate social media responses in real-time. Be interesting to see how this develops.
again, it will be one of those landmark moments (i.e. a moment that was not generally advertised) where those that were prepared would gain the most and those unaware, would, well - flounder! of course, you'd then have a proper FAKE news angle to tackle as it could (and would) be orchestrated both ways. in effect, a small(ish) window where it would be genuinely useful if you had the tools.
Post Reply

Return to “General discussion”