3
$\begingroup$

I'm want to do a PhD in NLP and I'm defining the topic at the moment.

I've heard that NLP can be used to track online sentiment which in turn can be used for algorithmic trading on stock markets. See this hedge fund

I was wondering if NLP techniques, maybe with online sentiment tracking, could be used to predict sport events on online betting sites? Would that be feasible? If yes, what research has been done on the subject?

$\endgroup$
  • 3
    $\begingroup$ NLP? Nonlinear Programming? Natural language processing? Neurolinguistic Programming? ... $\endgroup$ – jbowman Jan 11 '12 at 21:22
  • 2
    $\begingroup$ I favor nationwide loan processing :-). $\endgroup$ – whuber Jan 11 '12 at 22:24
  • 2
    $\begingroup$ @jbowman Given that the question is tagged [natural-language], I think you can infer with high confidence which one applies :) $\endgroup$ – Michael McGowan Jan 11 '12 at 22:42
  • 1
    $\begingroup$ @Michael McGowan - I am an idiot! :-) $\endgroup$ – jbowman Jan 11 '12 at 23:39
1
$\begingroup$

I believe it is feasible, but accuracy will be much lower then same thing with hedge fund. The key point about hedge fund research is that

For years investors have widely accepted that financial markets are driven by fear and greed

So, if people are afraid, they tend to sell, and if they are greed, they buy. Sentiment analysis just gives data how much people are currently afraid, and how much of them are greed, and thus predict market movements. (This is very rough model, of course, but you should get the idea).

Now about sports. Does fans' mood influence players in sport games? Well, most probably it makes some effect. But is this effect visible at all? I'm not certain about that.

You can also try to find correlation between fans' mood and game results (before the game starts, of course, after the end it makes no sense for prediction), but most probably it will be very similar to what bookmakers tell you.

Note, however, that I'm not sports fan, so I may miss some trends, that you are familiar with :)

$\endgroup$
  • $\begingroup$ I was thinking that perhaps the over confidence level of people talking about the upcoming match could reflect the odds of which team is gonna win. $\endgroup$ – siamii Jan 12 '12 at 9:58
  • $\begingroup$ @bizso09: as I said, this kind of correlation is possible, but I believe that people opinion reflects actual probability of the win of this or that team, so additional sentiment analysis doesn't seem to add much more info about match result. $\endgroup$ – ffriend Jan 12 '12 at 12:48
  • $\begingroup$ I don't think that the fans' mood affects the outcome so much as it may provide insight. So "opinion reflects actual probability" is correct (@ffriend). If these probabilities are somehow already known, then the sentiment analysis would be in fact be "additional", but how would one ever really "know" those probabilities? One could take their recent game history, weather conditions, the relationship between the players and the coach etc. etc. I think the mood of the fans could be another useful input. $\endgroup$ – Rohit Chatterjee Jun 28 '13 at 5:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.