2016 Semi-Final One Prediction: Our Infinitely Reliable Non-Scientific Model

As the 2016 Eurovision season got underway in December, information about artists, songs and presentation (or staging, if you prefer) have been dribbling out; throughout this period, the ESC fan community has developed its impressions of each entry. Many become convinced about who the winner will be after a national final or entry début. Once the Heads of Delegation meeting happens mid-March, we have heard each entry and know each performer.

However it’s not until we see each entry on the proper Eurovision stage that we have a clearer sense of each’s quality. Now, after a week of rehearsals, pop-up fan events, and various clips on social media, we now have some reliable data about the “total package” for each of 43 42 entries.

We have been aggregating all sorts of data to try and build a model that reliably predicts qualifiers from each semi-final.  In fact, we have worked on these sort of models for many years (long before 58points.com existed, on our personal blogs). But this is our first one for this site.

Before we share our prediction, we would like to explain how we came to our rankings.

Our Top Prediction Model

Whilst most Eurovision winners have a certain intangible je ne sais quoi, often many of the semi-final qualifiers can be predicted. While it is true that each entry is judged in two wholly different ways (professional juries versus the televoting public), but we think it remains possible to unpack key elements of the entrants in a semi-final and come up with a quasi-scientific quantification of each’s chances. We do this by consider each element a criterion: we make each element a criterion by quantifying different performance indicators (level of performance) for each. Over the years we have realized that none of these criteria should be over-weighted in relation to the others; some, in fact, are only worthy of a range of zero to one point.

Here is a description of our model for 2016:

Criterion Performance Indicators Notes
Bloc 0 or 1 Membership in a voting bloc. How well you do within the bloc is not consistent enough to quantify.
Diaspora 0, 1 or 2 Having no, a small but committed, or a large diaspora that can boost televote scores.
Jury & televote” bait” 0, 1, or 2 each Extent to which there are any elements aimed at either score components. Jury bait can be playing your own instruments, presenting yourself as a musician rather than “just a singer”. Televote bait can be showing some flesh, singing well whilst performing challenging choreography, flirting with the camera.
Performance 0, 1, 2, or 3 Extent to which act can sing well, and perform well for a TV audience.
Gimmick 0 or 1 A schtick to make the voters remember you. It need not be classy; in fact, it can be cringe worthy.
Established artists 0, 1 or 2 Act has an established, hit making career that gives them profile, usually internationally, but can be regional.
Selection 0 or 1 How entry was selected. Anything except “televote only” gets 1 points
Self-sung 0 or 1 Artist is also writer of song. Juries, in particular, love that.
Slot qualify Fraction A conversion of the percentage score for songs in this slot qualifying from a a semi-final.
Country qualify Fraction A conversion of the percentage for songs from this country qualifying from a a semi-final.
Prediction [scale] Poor – less than 4 points

Marginal – 4-5 points

Fair – 6-87 points

Good 8-9 points

Excellent 10 or more points


Here are the rankings, from most likely to quality to least. These are presented in bands, based on clusters of scores. The highest possible score would be 17 points. Two entries scored above 13.

Excellent: Russia (14.63 points), Armenia (13.63), Bosnia & Hercegovina (11.63), Iceland (10.41) and Croatia (10.94).

Good: Malta (9.94), Netherlands (9.88), Czechia (9.63), Cyprus (9.53), Hungary (8.96), Greece (8.63) and Estonia (8.05).

With five excellent entries and seven good ones, we need to cull two of the good’ns. Estonia is nearly a full point behind Hungary, so farewell Juri. Hungary’s score is a bit higher, while Greece has a better, more reliable diaspora. We are inclined to go with the former, making Hungary our 10th qualifier. That would mean Greece becomes the latest country to lose its perfect qualification record.

Limitations. And Outliers

The model is designed to reflect that in recent years both jury and televote scores in countries in voting bloc or with strong diasporas get points regardless of what they send. That’s not fair, but life isn’t fair. These sorts of models are relatively reliable in most years: most often they predict seven to nine qualifiers in each semi-final. They have never been 100 per cent accurate, however.

Where they have been almost useless is when there’s an outlier—an entry that stands out in so many ways that any disadvantages, such has not being in a voting bloc, having a poor track record qualifying from semi-finals and not seeming—at first—as the sort of entry that will inspire jurors or televoters.

Something like this:

Or this:

Both Austria and the Netherlands had poor records qualifying in the semi-final era prior to the 2014 Contest. Austria had only appeared in the top 5 of a Eurovision once since 1980, the Netherlands only thrice over the same period. Many thought Conchita’s entry was too marmite to be in the mix for victory; many also thought that The Common Linnets’ entry was ill-suited for the Eurovision because of its genre.

Maths can help us understand things better, but they cannot predict anything with 100 per cent accuracy. And sometimes the merit of a Eurovision is not easily quantified.

Click here for our full scoreboard for semi-final one. And stay tuned for our semi-final two prediction!