The Eurovision Song Contest 2026 is almost upon us – we’ll know this year’s winner in less than a fortnight.
To try to understand who the contenders might be at this year’s Grand Final, I’ve been using The Model to provide weekly updates from mid-March on ESC Insight, bringing you what it believes the Eurovision Song Contest could look like. The Model turns data like betting odds, community rankings, polls and even internal musical statistics into a single, numerical prediction of what could happen if the Eurovision Song Contest was held that same day.
This year, Finland has been The Model’s number one entry in all but the first week’s predictions. However, this week we’re learning more about each of the 35 competing songs than ever before. We’ll see previews of every song’s all-important staging concept, more bets on the Contest will be made, and projects like the Eurovision Audience Poll will give us greater insight into what we think fans and viewers will vote for during the three live shows.
Naturally, every prediction is just that – and on May 16, we’ll find out how wrong The Model will be. I say how wrong because trying to predict the scoring of 35 national juries, 34 countries’ viewing public, what the EBU will do with San Marino’s ‘televote’, and a single set of 12 points given by the Rest of the World, is an impossible task. While I always believe The Model has every chance of getting the outcome of Eurovision correct, no tool can see the future.
So, as we prepare to head to Vienna, let’s look at ten realistic things The Model may miss at the Eurovision Song Contest 2026.
1: Eurovision fans and Eurovision viewers want different things
Any update I do with The Model is based on the data we have available at that given moment. At this point in the process, with the Contest weeks away, the vast majority of those engaging with polls and rankings are those who are deeply engaged with the Eurovision Song Contest all year round: like you and me. Our opinions may be based on greater knowledge of the Contest and what is likely to do well, but these often create an echo chamber.
Additionally, bookmakers will look at the data they have available to them when pricing their markets for the Contest. That means, once again, the opinions of fans can look much firmer than they actually are in reality. The Model itself latches onto all of this data and creates a most likely outcome – but what it’s working from is almost entirely crafted by a small, passionate subset of the Eurovision viewing public.
That means, even for the semi-finals, there’s an influx of new viewers who are coming in completely cold – and what the viewing public latches onto is becoming less predictable. In 2025’s second semi-final, the bookmakers surprisingly predicted that ‘Bur Man Laimi’ from Latvia would fail to make the final in Basel. While many fans, and The Model, accurately called its progress, they completely failed to see it coming in second place in that show.
And if that influx wasn’t enough, ten times the audience for the semi-finals on TV only tune in for the Grand Final. While not all of them vote, of course, that’s another borderline reset of the narrative. This is why often semi-final results and Grand Final results can be totally contradictory – and because we don’t know the semi-final results on the night, it is very difficult to calculate.
2: The New-Look Juries
This year, the juries are getting their biggest shake-up in a generation.
The groups casting 50% of the points (in all shows for the first time since 2022) will be bigger, expanding from five members to seven. Not only that, two of those jurors must be 25 or under, demonstrating that Eurovision wants its results to more accurately reflect the tastes of its target demographic. This is the first time we’ve ever seen this, and that means there is guesswork involved.
On one hand, there is a belief among some that introducing more jurors and reducing the average age overall may create more consensus between juries and the public. After all, the public’s desired winner hasn’t walked away with the crystal microphone in three years, and there is a risk of fewer people voting if they feel like their song will be dismissed by the jurors before their points are even added.
But conversely, there’s another school of thought that says juries could actually go the other way. As Ben Robertson said in his recent study of the new jury rules, “the biggest statistical factor we have observed within Eurovision juror taste hasn’t been age, gender or other variables, but who one watches the Song Contest with.” Because of this, you could find jury opinions solidify further on the scoresheet as the average of seven creates a firmer mean than the average of five. In a year that so far appears to have no runaway favourites, how these new juries behave will be critical to seeing how Eurovision goes.
3: The Model’s Story Isn’t What the Viewers Want
In 2025, The Model ran several updates in the lead-up to Eurovision that painted a clear picture of the Contest to come.
It said that Sweden’s KAJ were the favourites with ‘Bara Bada Bastu’, just like the bookmakers, but their issue could be in the juries. On that side, France’s Louane with ‘Maman’ was projected throughout as the song that could win the most points from the jurors, with Austria seen as a solid bet on both sides, but not enough to cross the winning hurdle.
That narrative wasn’t completely off – however, it’s story that year didn’t foresee two key things that shifted the race in JJ’s direction.
France’s staging, which was much-anticipated in the lead-up to the Contest, ended up almost harming Louane more than helping her, with Graham Norton’s famous comment that it looked like something other than sand an indicator of the general reaction around Europe. Jurors also didn’t support the song as much as was expected. That meant Austria, the second favourite with The Model on the jury side all the way that year, was able to take a lot of those points France ‘lost’.
The second force was that Sweden wasn’t as powerful in the public vote, and that Estonia was going to be the song that the public preferred. As the two countries were in the same voting bloc, and had songs that appealed to a similar audience, Tommy Cash’s points almost came directly from Sweden’s expected televote. But with Estonia struggling to gain jury points (as was predicted), ‘Espresso Macchiato’ still didn’t have enough to take the victory.
By understanding those two things, it would be easier to predict Austria’s victory. This year’s story, with Finland seen as the favourite on the public vote side and benefitting from no runaway leader on the jury side, might be accurate. But the audience might have different ideas to what we have of this year’s narrative.
4: Unexpected Momentum
The biggest blindspot in trying to predict the Eurovision Song Contest is that almost nobody has heard the songs yet.
Of course, we fans have. We’ve lived with these songs for the best part of three months, four months in some cases. We know which songs the fans largely like and don’t, and that colours our understanding of how each song should perform at the Contest. But there’s a huge difference between hearing the song in its national final, or even at a pre-party, and seeing it on the Eurovision stage.
When the public sees these songs, they won’t be looking at them in anything like the same way we do. Most viewers are watching for far less active reasons – a friend has invited them to a party, they were channel-hopping, or they have it on in the background while watching another screen. As our conversations are only with actively engaged fans like us, it’s hard to know which songs catch somebody totally off-guard in the way big contenders do.
While there is evidence of the viewing public broadly becoming more knowledgeable about the entries going into the Contest, there are reactions we simply can’t predict. For example, Sal Da Vinci was nowhere near the conversation for winning the Festival di Sanremo. But, when people heard ‘Per Sempre Si’ and felt a warm, fuzzy feeling inside, suddenly the whole town moved to Sal’s rhythm. Momentum can come from anywhere, and totally rip up the script.
5: The High Margin of Error
Many statistical models are based on running multiple simulations of something and creating an average. For example, the Opta Supercomputer, which predicts the outcome of the final table in England’s Premier League football competition, runs 10,000 simulations of all matches to try and calculate the most likely scenario.
The Model runs slightly less for each forecast of the Eurovision Song Contest. One.
Now, compared to 10,000, that might seem pretty primitive. And in some ways it is – but running one simulation presents a far more accurate version of the Contest than 10,000 simulations can when averaged out. I’m not saying either approach is better or worse – but it creates a different outcome. Running 10,000 simulations, for example, vastly decreases your margin of error, which is great if you want to work out the probability of an outcome.
With The Model, I want to know what the outcome will be. No ifs, no buts.
That means that, if the underlying data is accurate, The Model should be far more accurate in predicting the level of success of each song. But if the data going in doesn’t match up with the votes on the night, the level of difference can often be much higher. It’s a higher-risk approach that could be almost exactly right on some things, but totally wrong on others.
I would say that, in The Model, it’s fair to look at the numbers with a margin of error of around 15% in the semi-finals, and 25% in the final. If every country comes in within that scale, from my perspective, I’d be delighted!
6: The Rise of…English Language Songs?
This might sound weird. But go with me on this.
The major shift in the Eurovision Song Contest in the last decade has been the proliferation of entries in native or other languages than English. A decade ago, at the 2016 Contest, all but one of the songs in the Grand Final was performed with at least some English lyrics. In current projections, it’s highly likely that this year’s final will have more songs in a foreign language than English overall.
However, English still remains Europe’s most commonly-spoken language. And, rather than songs in native languages standing out among a majority English line-up, this year the inverse will be true. In 2025, we also had a similar phenomenon – the majority of songs who overperformed compared to initial expectations were songs with English lyrics. Despite the unmistakable trend towards native language entries, the last three winners of the Contest were all jury winners with entirely English lyrics.
This year, in particular, the main contenders are in native languages like Finnish, French, Danish and Greek. So I wouldn’t be surprised if songs like ‘Eclipse’ from Australia, ‘My System’ from Sweden and others stand out a little more to viewers who may like the sound of those other contenders, but just don’t understand what they’re about. The Model doesn’t take song language into account – usually the correlation is low – but in extreme circumstances like these it could matter.
It’s a powerful force that has shaped Eurovision in the past. It was so powerful in the 1990’s the UK and Ireland were actually good.
7: The Contenders Themselves
In The Model this year, I quickly identified what I saw as eight songs that looked like they could be in some form of contention to win Eurovision this year. Those were: Finland, Denmark, Greece, France, Sweden, Australia, Ukraine and Israel.
But that is based on a small sample size, mostly made up of those who love the Contest. While fans have been firmer than usual around Finland in particular, with ‘Liekinheitin’ currently dominating the Eurovision World poll which often gives a good indication of a song’s chances, bookmakers haven’t had the same kind of confidence that they had with KAJ and Baby Lasagna (incorrectly as it turned out) in the previous two years.
A lot of the justifications for those songs being contenders rely on previous assumptions about Eurovision holding true. Israel and Ukraine, for example, are heavily predicated on their public vote being in line with previous years. Denmark’s is based on its strong staging at the national final and whether it will still be a standout once everyone’s brought their concept to life. Even Finland’s pre-Contest favourite status is based on the opinions of a small segment of the audience.
So, the protagonists of our story now may not be the protagonists of the Song Contest. And there is every chance that a song like Moldova or Italy could come out of nowhere to become a serious contender for the crown.
8: The Anti-Eurovision Effect
To predict the Eurovision Song Contest, you need to know the rules. Ballads usually appeal more to the juries. High-energy songs usually appeal more to the public. A more dynamic staging is more likely to catch the eye than someone standing there and singing.
All rules are made to be broken. This year, there’s a song that looks to be deliberately trying to be as non-Eurovision as possible.
In The Model, Portugal has almost never done well. In 2025, it was very negative on ‘Deslocado’ by NAPA, a song that everyone was lining up to declare ‘dead-on-arrival’ before the first semi-final that year. After it qualified, suddenly everybody claimed that it was because the song was a breath of fresh air. What they really meant was that it flew in the face of the rules – sometimes, quiet and tender is the loudest sound on stage.
This is not a one-off. In 2024, ‘Hollow’ by Dons was similarly written off by The Model, only to qualify for the Grand Final comfortably. Portugal’s winning song in 2017, ‘Amar Pelos Dois’, was nothing like a pre-Contest favourite either – and it won the most points of any song in the Contest’s history. And this year, ‘Rosa’ by Bandidos do Cante appears to be going on the same trajectory, with The Model not seeing a path for it to qualify.
But, with the group standing there in plain clothes in the first rehearsal photos, they look to be doing something smart. In such a dynamic first half, with Finland, Greece and Sweden all loud, high-production efforts, Portugal’s will be as far from Eurovision as it gets. And, for a large group of audience members who don’t care about the rules we think of this Contest following, that’s enough to completely shatter convention.
9: The Unexpectables
The Eurovision Song Contest is live. And when anything is live, nothing is expected.
The one thing that no simulation can ever predict is what happens in the moment. Pressure does things to even the best performers – and the idea of millions of viewers changes the context of any performance. Coupled with the fact that the lead vocalist doesn’t get to rely on their studio vocals, and every performance becomes a high-wire act.
The Model can account for everything right up until the performance – but not a bum note. Nor can it foresee the tech that each Eurovision performance relies on letting them down in the crucial moment. Somebody stepping on the wrong cable at the wrong time can derail a performance’s key shot. All of those things, naturally, have a massive impact on the impression each viewer gets. And with most viewing the performance green, the worst thing they can think is “ooh, that wasn’t quite in tune”.
On Inside The Model, we recently spoke to professional vocal coach Zoe Stibi about how the singers manage their voices ahead of the Contest. In a competition where almost everything else is controlled and tight, the voice is the distinctive live element. That makes it the most important – and most risky – part of every performance.
10: The Future is Unwritten
Tell me this – if The Model was absolutely, 100% guaranteed to be right about everything – would you want to look?
I’d bet your answer is no. Because the whole thing that makes trying to call the winner of the Eurovision Song Contest so exciting is that the future is unwritten. We cannot possibly account for all of the things that will affect the result next weekend – the weather on the day, what is being reported in the day’s news, whether one of the artists had a bad night’s sleep or not.
I created The Model to try and get as close as possible. So close you can almost taste it. Release your inhibitions. FEEL THE RAIN ON YOUR SKIN…wait.
And I do believe that, next week, The Model will be able to help us predict each show with greater accuracy and knowledge than just by looking at the betting odds or talking to your friends. Last year, it successfully called Cyprus and Czechia’s surprise exits at the surprise semi-final stage, while also spotting Iceland and Armenia’s qualifications when almost nobody else made the call.
There will be moments like that – and there will inevitably be misses. Even as we have more data about how Eurovision works than ever before, it’s almost becoming harder to see the valuable data through all the noise. My hope is that this year The Model, more than anything, simply allows us to think about what’s to come through a more accurate, analytical lens.
We’ll see if it gets things right. But it’ll be even more exciting to see what it gets wrong.