Eurovision betting – lies, damned lies and statistics

Eurovision season has well and truly begun: this weekend Malta will add its selection to the six we have already. We’re yet to hear the final version of some, and I haven’t yet felt the need to write a lengthy, pre-rehearsal analysis for any of them.

But like other Eurovision punters, I’m already making calculations. The semi-final allocation has taken place, and the good people at Esctoday.com came up with a predicted scoreboard for each heat based on previous semi-final voting.

The beauty of betting on Eurovision is the wealth of statistics and the need to carefully interpret them. I’m going to use this well-intentioned Esctoday forecast as an example of how to approach such stats, and an interpretation I’m going to focus on is the concept of ‘hard points’ and ‘soft points’.

Bear with me, I’ve made up these terms myself.

Esctoday is open about some of the difficulties with their numbers. Points given in the final were not taken into account. Had they been, it might have helped fill in the gaps: 14 pairs of countries will meet each other in the semi-finals for the first time; and the scoreboard is not full because in some cases there weren’t ten countries to give points to as a result. For example, in the second semi scoreboard, Austria only gives out a 12, 10, 8, 6, 4 and 2.

To be fair, there is a reason why it might be useful to focus on semi-final votes alone. In some cases the audience is much reduced and voting patterns thus slightly different. For example, in the United Kingdom, where it’s watched midweek on BBC3 by a far smaller amount of people than the Saturday night final, migrant groups have a bigger impact in the semi-final televote.

Concentrating on semi-final votes alone isn’t the main concern of this article, however. It’s a more general warning about taking stats such as these at face value. One only has to compare Israel’s highest scores in this 2014 second semi-final chart with its points from those countries when failing to qualify in 2013: the prediction is 10 from Finland (in reality, 2 last year), 8 from Georgia (5 last year), 7 points from Norway (0 last year).

This is where my idea of ‘soft points’ and ‘hard points’ comes into it. The concept is that ‘soft points’ are given for song strength and performance alone, rather than the more concrete voting alliances which create ‘hard points’. Soft points come and go; hard points are more likely to reoccur year-after-year, no matter what the performance.

Norway provides a good example. It is fourth, one place above Israel, in Esctoday’s second heat scoreboard, based on excellent previous showings from the likes of Maria, Alexander Rybak and Margaret Berger – strong entries that received plenty of ‘soft points’ across the board.

But weak vocal performances from Stella Mwangi and Tooji indicate how many ‘hard points’ Norway can rely on in tough times, and the answer is: not many. Tooji squeaked through thanks to a 10 from Sweden and an 8 from Estonia; Stella failed to despite a 10 from Iceland and an 8 from Finland.

Finding a country’s weakest previous entries and looking at which allies, if any, still gave it points, has been an invaluable tool for me. These are the ‘hardest’ points of all. Of the four that stood by Norway in its flattest moments, only Finland is in its semi-final. Send a poor performer to Copenhagen, and Norway will struggle to qualify, let alone manage fourth.

In one sense, the growing power of the jury vote eats into this concept of ‘hard points’. In my previous article, the example I used indicated Spain’s points for Romania were no longer ‘hard’ based on the five-person jury’s power to put a country at or near the bottom of its rankings. (This ability to rank and thus treat an entry negatively is not open to televoters – who merely register a positive vote.)

Juries thus have the potential to soften any ‘hard points’, but some are more enthusiastic than others. One reason for the Balkan wipeout in last year’s semi-finals was a lack of jury love from some televoting allies. In semi 1, the Slovenian jury negated the televoting power of Serbia (given 5 points), Croatia (2 points) and Montenegro (0). In semi 2, the Swiss jury did the same with Albania (5) and Macedonia (4).

In Esctoday’s historical study, Macedonia is bottom in heat two despite 12 points from both Slovenia and Switzerland. I’ve just indicated that the Slovenian and Swiss panels show no loyalty to televoting allies. Therefore, Macedonia has no significant hard points in this year’s semi at all. It’s in an even more difficult position than indicated.

In other cases last year, ‘hard points’ held up more robustly depite the new system. In the final, Georgia’s average jury ranking was middling and its average televote ranking was very low. But it still managed decent totals on the leaderboard from its traditional allies (Azerbaijan, Armenia, Lithuania, Ukraine, Russia and Greece), where presumably its jury rankings held up much better, even in a field of 26.

We can use these ideas and return to the Esctoday semi-final tables. There’s no way that Iceland has more ‘hard points’ than Ukraine, yet it comes out one point ahead in the first semi-final chart based on past performance.

In the 2008-13 semis, Iceland scored 68, 174, 123, 100, 75 and 72 points respectively, due to performance rather than voting allies – Iceland is geographically remote and doesn’t have many. Ukraine has had plenty of memorable performances during that era too, but even when its televote score was negligible in 2012, its jury vote held up where it mattered. Therefore, I have more confidence in Ukraine’s score being higher should both send similar quality entries.

There is a big caveat for 2014. A tightening of the rules this year, with more announced yesterday (putting the burden of sanction on individual broadcasters if irregularities are uncovered, even if no direct link can be found), may affect certain recently established patterns in televoting and jury scores. Punters should be aware of where anomalies have existed and be warned – these can change quickly.

But the caveat reiterates the central message of this article: statistics only take you so far when punting on Eurovision. What counts is looking at entries and scoring patterns on a case-by-case basis and not taking too much for granted. Let us know your thoughts on these and any other Eurovision-related points below.

11 comments to Eurovision betting – lies, damned lies and statistics

  • Ross Taylor

    Great Article yet again.

  • Ben Cook

    Spot on Daniel. I always find this statistical projections to be meaningless. Just because a country has sent strong entries for the last 5 years and qualified every time (e.g. Iceland), doesn’t mean they will do so again. Having heard their NF songs and noting that they are in an extremely tough first half of the first semi, I don’t think they are going through.

    • Whoever

      “Just because a country has sent strong entries for the last 5 years and qualified every time (e.g. Iceland), doesn’t mean they will do so again.”

      To build on this you can read Bertrand Russell’s Inductivist Turkey, a very short story with a very interesting point on how the repeated occurrence of an event in the past does not guarantee its continuity over time.

  • I like this terminology of soft points and hard points. Simplifies things a bit. 🙂 Very interesting read. Right now I’m mainly wondering if I should trust the EBU’s claim of past legitimacy and use an Azerbaijan top 5 or 10 as my super safe foundation bet, or will the new announcement cause an upset? I suppose their song would help, wouldn’t it? Hahaha.

    • Chris Bellis

      I always thought that people on this board were well educated (quite contrary to the general impression of betting folk, especially those who bet on TV events, but I never thought I would read about Bertrand Russell here. It will be Wittgenstein next.

      • Whoever

        Well, block voting has a lot of politics, but it also has something to do with worldviews and linguistic intelligibility… The limits of my language (or their near relatives) are the limits of my voting? 😉

  • An off-topic point here, but here’s a little nugget that I feel I’m going to be working into my criteria when looking for something that’s going to do well, one which I think might offer some chance for reflection on previous contests.

    I am now inclined to listen out for slow-building ballads, regardless of country, and presume that they may receive a lukewarm televote. I only need to refer to Quedate Conmigo and Georgia’s Waterfall to say that these kinds of songs generate fan hype because they have drama and emotional uplift which sets the pink vote off like a Catherine wheel. These are theatrical songs that work in the very typical Eurovision context, and yet they can only manage middling results despite performances that are hard to fault.

    The reason for this, I think, is that they simply aren’t instant enough as songs. Ok, so they may well create an impression as their final chorus bursts into life and Pastora belts it like her life depended on it, but is that going to be as effective as a consistent strong, catchy chorus, either in upbeat or ballad form, when the phone lines open? The evidence suggests not. Infact, even though the voting systems were different in 2012 and 2013, to compare a Spanish 10th place to a Georgian 14th(?) despite their voting power and with songs that are structurally and stylistically similar within the ballad genre, Spain’s result in 2012 is actually quite a remarkable achievement for them.

    My premise in 2013 for being quite sure of Georgia doing well were comparisons to In A Moment Like This and Running Scared. In other words, I believed schmaltzy hetero love songs to be some kind of overreaching exception that vacuumed up points across the board, but with this theory, I now realise that Georgia was the black sheep of these three examples simply because it was a slow building, dramatic ballad, whereas Denmark and Azerbaijan’s entries had a consistent, catchy chorus. Denmark in particular reminding many people of two iconic 80s songs by The Police and Tina Turner.

    I think we can all save ourselves a few headaches by learning to distinguish between types of ballad, and perhaps in the future, types of uptempo songs. At the moment, they seem to be referred to, even in the ESC betting world, as very polarised genres.

    Would like to hear other people’s thoughts on the importance of song structure when winning over votes. Particularly televotes.

  • john kef

    I have done a statistical analysis on the past 12 double-semifinals concerning the least possible points needed for a song to qualify. I gave an extra attention to the last 8 semifinals where the 50-50 jury-televote system runs and the magic nymber is 22.50% of all the points an entry could get.

    Only Sweden in 2010 didn’t make it with a 28.7%- 62 pts
    and Malta, Armenia in 2011 scoring exactly 22,5%-54 pts losing by a single point from Switzerland’s 22.91%-55 pts.

    4 countries qualified with less than 22.5%
    a. Moldova 2011, 21,43%-54 pts
    b. Hungary 2012, 21,67%- 52 pts
    c. Norway 2012, 18.75%- 45 pts
    d. FYR Macedonia 2012, 22.08%- 53 pts

    Calculating the number of countries voting in this year’s semifinals i believe that the margins are:

    1st Semi- 49 pts
    2nd Semi- 46 pts

    A country might qualify with even less points but i can’t see a country being disqualified scoring those points.

    I have also counted the songs qualified from the 1st and 2nd half of the semis and the total score is

    1st half: 51
    2nd half: 69

    The 1st half’s best score is 5-5 (several times) meaning that we never had so far more songs qualifying from the 1st half.
    the 2nd half’s best score is 7-3 (2008a, 2009a, 2009b).

    * The things are a little bit different in the semis with 17 participants or less. The songs are not that easily forgotten with a total score 1st Half: 19 – 2nd Half: 21

    I have also counted the songs that qualified appearing in the last 6 places of appearance order. The score there is

    Q: 54
    NQ: 18

    I hope that these numbers might help you a bit.

    • Daniel

      Thanks John, these are great stats to bear in mind. Have also added a link in the article because the Slovenian jury did reveal their top ten rankings after last year’s contest. There was no room for any of the three other Balkan countries in their semi-final top ten, among the 15 countries to rank.

  • I would dare to go even further, Daniel. You mention Georgia having ‘hard’ points even from the jury, based on 2013 results. But last year Georgia sent a song in a genre juries ALWAYS vote for, regardless the country. These ‘big old-fashioned ballads’ always score heavily with the juries. Same goes for Ukraine 2012: Gaitana may have had a weak song, but her vocals were great, and it is common knowledge juries pay specific attention to vocal qualities (more than musical/song qualities). So even there, I would say: nothing is certain.

    • Daniel

      Hi eurovisie, and thanks for the excellent comment. You are right in saying that the Georgian song was the kind of thing juries often go for, as indeed were Gaitana’s vocals.

      But we have on public record the jury results of 7 countries, none from the ex-USSR bloc. Only one of those juries put Georgia in their top ten (the UK placed it 7th). Two placed it very low indeed (20th with the Dutch jury and 22nd with the Swedish jury).

      What the new system emphasises is whether juries reinforce the televoting alliances of their particular nation. I think this was the case with Georgia 2013, which scored well with all its televote allies.

 Leave a reply...