Well, there is nothing like predicting the 1-2-3 of something in the correct order to put the pressure on the next time you have to forecast the same thing. However, last year’s Eurovision Song Contest projection included the use of the bookmakers’ odds which I regard as at least partial cheating. So, this year I wanted to avoid bookmakers and use a number of objective methods. These methods, as you can see if you read on, give us a 1-2-3 this evening of Russia, Ukraine and Australia.
Just like last year, the televoting part of the Eurovision Song Contest has been produced by my friend and colleague @oneminutecoach whose database of the history of the competition and its voting patterns provides an entirely objective view of how countries tend to vote for one another. However, although this approach provides a decent starting point, it cannot be the whole story as there is also a quality element at play. Certain countries may tend to vote for one another but if the quality of the song or performance are not high enough, those countries benefitting from this cannot possibly win or even perhaps do well.
In order to estimate the jury side of things, we need to have some measure of the quality of the performance and song. In order to do this, I have recorded the number of YouTube views of the songs from 24 hours before the start of the competition until now. That time period was chosen to attempt to avoid the bias introduced by the semi-finals being two days apart. However, there is still that risk although the top-10 does seem to be reasonably well distributed with acts from each of the two semis. I have used the videos of the performances on those two evenings, which included the six songs which qualified directly to the final. However, in hindsight it may have been better to use the official videos as these are linked directly from the running order on the Eurovision 2016 website.
During last year’s competition I was introduced to the excellent @kitlovelace on Twitter who comments on the songs’ strengths and weaknesses based on their key, tempo, the presence of any key changes, the time signature and the language they are sung in. He has a database of this for all of the songs this century and writes about the contest for Popbitch (you can download his guide to tonight’s competition here: http://popbitch.com/home/2016/05/10/the-popbitch-guide-to-eurovision-2016/). This is what he has to say on tonight’s hot favourite with the bookmakers’ Russia for example:
Russia – the hot favourite – stumbles on a few issues.
1/ It has a key change in it and there hasn’t been a winner to use a key change in nearly a decade now (last one was Molitva in 2007).
2/ It is also (unless I’ve really cocked this up) timed at 128bpm, which has been the tempo marking of three of the last five entries to come dead last. (Six songs that came dead last since 2000 have been 127/128bpm)
3/ It’s also in a compound time signature which, while not a disaster, is very rare (winners and losers since 2000 have been written exclusively in 4/4). However, that time signature might be the thing that saves it from the 128bpm trap though because the triplet effect makes it sound faster than it is.
So, having collected all of this information, we can now make a projection for this evening without any recourse to bookmakers’ odds, other than to see if our forecast makes sense. Last year we weighted the voting behaviour at 50% and the “quality” also at 50%. I’m going to drop the quality measure to 40% and weight Kit’s trends as the remaining 10% of the total. Below is the full ranking resulting from our data driven projection.
So, despite the song’s potential issues, it looks like this evening’s winner will be Russia with Ukraine second and Australia third. Poland is in fifth position because of its enormous popularity on YouTube as neither voting behaviour nor musical trends rates it. France is fifth favourite at the bookies but comes in at number 23 on our method which makes me concerned about its validity. I’ll be back once the result is known to again assess how we did and what we can learn for next year.