Eurovision statistics: final predictions | Hacker News Reader

Eurovision statistics: final predictions

mewo2.github.com

33 pointsdlib11 comments

From what I've read, Loreen from Sweden was pretty nervous during her performance (in comparison to the local competition, where she seemed very secure) which affected how she acted and sung on stage pretty badly. It's nice to see that she still appears to be a favourite, and I hope she gains some courage for the final.

(I am Swedish, I have yet to watch this semi-final, and personally don't enjoy her song that much, but is still happy to see that we're doing well.)

I wonder if there might be a difference between viewers of both semi-finals. People have more intent to watch the semi-final where their own country participates.

Also, some countries will have failed to advance in the first semi-final. Would they be even less likely to watch the second semi-final and the final?

If there would be a way to estimate the viewer's distribution for each show, it could improve the accuracy under this model.

Countries only vote in the semifinal in which they qualify (except the automatic qualifiers, who are assigned a semifinal to vote in). There probably is a systematic difference between the semifinals and the finals though.

Yesterday before the show I made a bet with a return of 2.0 that a participant of Semi 2 will win the contest (I think it will be Sweden, but wanted to hedge it somewhat). So it is nice that your calculations predict high chances for Sweden, Malta, Norway and Estonia. ;)

Comparing the MCMC calculations and book maker odds are interesting. For example you rank Turkey (which I think is the better song) higher than Azerbaijan, but in the odds it is dramatically different. A return of 3.0 when Turkey comes ahead. Hmhmhm.

Be somewhat wary of the predictions for the countries which have automatically qualified - they're based on much less information than the other countries. The reason that Turkey is ranked above Azerbaijan is that the model has absolutely no information about the quality of Azerbaijan's song. Turkey, on the other hand, made it through a semifinal, so their song can't be completely abysmal. That said, Turkey do have a pretty strong structural advantage in terms of incoming vote patterns. Out of the 10,000 simulations, Turkey beat Azerbaijan about 75% of the time.

Also, this should go without saying, but this is not financial advice, please don't blame me if you lose money betting on the predictions of this model, etc.

I am not a big gambler. But I wonder if these sort of calculations could be better in forecasting than prediction markets like intrade.com.

Very interesting work on the Eurovision at least! There are so many potential variables. Age, gender, BPM of the song, changes in the performance, political stuff. For example I don't expect any points for Azerbaijan from Germany, because the media reported very critically on the undemocratic regime. But don't know how this will play out in the rest of Europe. Are you American? This song contest and its history must be pretty strange for outsiders.

I'm Irish, so I'm fairly familiar with the contest's history. I do live in the USA now though.

I am a bit skeptical about how effective politically motivated boycotts can be in Eurovision. The problem is that it's not possible to cast a vote against a country, so the only way a boycott can work is if it convinces a lot of people who were otherwise going to vote to abstain. That said, there has been quite a lot of coverage in at least the Western European media which has been very critical of Azerbaijan, so we'll see. It's also just possible that nobody will like their song - I'm not sure how we'd tell which had happened.

Ah, damn, that was stupid! :-P At least it seems like Sweden is winning.

What are twitter and the book makers predicting?

Sweden are pretty clear favourites at about 5/4, followed by Russia (7/1) and Italy (12/1): http://beta.betfair.com/special-bets/market?id=1.104582460

You can host blogs on github?

Reply on news.ycombinator.com