2016 forecast: final
Image source: Andres Putting (EBU)

2016 forecast: final

There were wolves! And holograms! And a naked man! And somehow Belarus failed to qualify for the final.

A slightly less surprising set of qualifiers this time around. Israel and Latvia were both only slightly less likely than Albania and Macedonia, so that swap is the sort of thing we should expect to happen.

More unusual is that Denmark and Norway both failed to qualify, which means than no Scandinavian countries have made it through the final (apart from Sweden, who automatically qualify). This is definitely evidence in favour of the hypothesis that the voting this year is a little bit weird by historical standards, possibly due to the new system. I’m definitely interested to see the voting breakdowns for this.

Final predictions

Winning probabilities for 2017 final

This is an interesting one. Normally I’d say that having all of the other Scandinavians knocked out (but still voting) was good for Sweden, and the model agrees. However, if there really is something going on with the voting which is hurting the Scandinavian countries, then that would presumably affect Sweden as well. I’m not sure I trust the boost that they’ve got here, but it can’t be denied that they have excellent form.

The next group of four countries are all in a statistical dead heat. Russia might be hurt a little by there being a wide array of ex-Soviet countries present, confusing the voting from that region. Ukraine probably suffer from the same problem, but to a slightly lesser degree. Australia have moved back up—they’ll probably capture some stray votes from countries that like Scandinavian-style pop, but have nobody left to vote for. As for Austria, the model is still bullish on them, but their odds with the bookies have lengthened considerably (although I’m not quite sure why).

Douze points

Let’s look at the twelves:

Jury 12pt probabilities for 2016 final

I’ve plotted here the probability of each country’s jury giving twelve points to each performer in the final. The televote probabilities are indistinguishable, so I haven’t duplicated them.

We’re missing a few of the usual suspects in the final (Greece, Romania, the aforementioned Scandinavians), so we’re not going to see a lot of the usual relationships. Greece will probably give 12 to Cyprus, but this relationship is a lot weaker than its counterpart in the other direction. It’s not clear what Sweden will do with its votes—Australia is the most likely destination according to the model.

The big winner here is Serbia: three of the six most likely twelves go to them, from Montenegro, Croatia and Macedonia. In general, the top ten are:

  • Greece → Cyprus (52%)
  • Montenegro → Serbia (48%)
  • Bosnia-Herzegovnia → Croatia (45%)
  • Croatia → Serbia (40%)
  • Albania → Italy (37%)
  • Macedonia → Serbia (37%)
  • Belarus → Russia (35%)
  • Denmark → Sweden (34%)
  • Sweden → Australia (34%)
  • Azerbaijan → Ukraine (33%)

The eagle-eyed will notice that these numbers are a lot lower than in previous years—the Greece-Cyprus link is typically more like 80 to 90%. The reason they’re so low is the new voting system, which introduces a lot of variability at the level of individual votes (although it all seems to level out once when added together).


That brings us on to bellwethers: which countries are good predictors of the winner? Or, in the new system, which countries’ juries are good predictors? (televotes will be announced at the end, so they’re not much use for prediction)

Bellwether probabilities for 2016 final

The answer, it turns out, is nobody. Hungary, as usual, have the best claim (despite getting it wrong three years running), but they still barely have a 20% chance of picking the winner with their jury selection.

The problem is the voting system again. Even once we have the winner of the jury vote, that can be completely overturned by the televote. In fact, in the model simulations, the country which is leading after the jury vote goes on to win the contest only 48% of the time. Even with a lead of over 100 points, that number only increases to 60%. Anything the jury can do, the televotes can undo just as easily.

So the upshot of all this is that the contest is less predictable than ever. Be very wary if a jury-friendly song is leading after the jury votes. And don’t trust the Hungarians.