EUROCASH #1: Introducing an alternative way to predict the outcome of Euro 2024
Every day until the start of the Euros, I'll use a model with a winning track record to predict outcomes - here's how it works
Euro 2024 kicks off a week today in Munich, when hosts Germany play a tournament opener against Scotland in the Allianz Arena, the home ground of six members of the German squad who play for Bayern Munich.
Over the next eight days, I’m going to try to predict how the event will unfold, group by group, using a model based on the insurable value of the players involved.
It’s a refined - and hopefully improved - version of a model that has worked before, predicting the winners at two World Cups (in 2014 and 2018), and wasn’t far off in a third, more of which in a moment. The model didn’t pick favourites on any of those occasions.
Between June 14 and July 14, the 24 nations at Euro 2024 will play a total of 51 matches across 10 venues. According to bookmakers, England will start as marginal favourites for the title, just ahead of France, with hosts Germany not far behind in third.
There’s then a gap to Portugal and Spain, who are neck and neck as fourth / fifth favourites, and a further gap before Italy, the Netherlands and Belgium.
Croatia and Denmark are the shortest priced outsiders, both at around 40-1 to win (stake one pound / euro / dollar to win 40). The rank outsiders, at prices from 500-1 to 750-1, include Slovakia, Georgia and Albania.
I don’t think an outsider will win. The model doesn’t think an outsider will win. The bookies don’t think an outsider will win. The world rankings don’t think an outsider will win. And Sporting Intelligence’s resident pundit in this Euros preview series, running from tomorrow to next Friday, doesn’t think an outsider will win either.
That pundit is Thomas Hitzlsperger, aka Der Hammer, who spent five years at Aston Villa, won the Bundesliga with Stuttgart, and also played at Lazio, West Ham, Wolfsburg and Everton. He was capped 52 times by Germany and reached the Euro 2008 final with his country. We’ll hear more from Thomas from tomorrow onwards.
While neither I nor Thomas nor the bookies nor the rankings expect an outsider win, remarkable outsider wins have happened in the European Championship. Denmark won in 1992, in a tournament they failed to qualify for! Having gained late entry because of Yugoslavia’s disqualification due to the war there, they started that eight-team event as 20-1 outsiders.
At Euro 2004, outsiders of a much greater magnitude prevailed when hosts Greece, who began the tournament at 150-1, beat Portugal in the final.
The Euros have also thrown up some stunning individual wins, and against-the-odds runs deep into the knockout stages. At Euro 2016, Iceland beat England in the last 16 (Iceland were 14-1 outsiders to win that match). At the same event, 80-1 outsiders Wales reached the semi-finals before losing to eventual champions Portugal.
Anyway, back to the model, and its genesis.
In Spring 2014, I was contacted by an analyst at the Centre for Economics and Business Research (CEBR) in London, asking if Sporting Intelligence could provide some salary data for England footballers who might play at that summer’s World Cup in Brazil.
I’d just published the fifth edition of the Global Sports Salaries Survey (GSSS), which ESPN The Magazine also published, in a deal where they paid Sporting Intelligence an annual fee for, in effect, first rights to reproduce my GSSS research worldwide each year. We worked together in 2012, 2013, 2014 and 2015 before I collaborated with other partners from 2016 onwards.
ESPN The Magazine was a substantial, high-quality glossy selling 2.1m hard copies a month but would cease to exist as a physical product in 2019.
The CEBR wanted access to my data, the data that underpinned the GSSS, and they quickly went from wanting just the England data to data for all 32 competing countries in the 2014 World Cup. Sure, I said, I’ve got most of it already, for something like 70% of the 736 players who will be in Brazil, because they’re in my database. Give me time and I can pull together the rest.
My CEBR contact then revealed to me that his client was actually Lloyd’s of London, the global insurance underwriter. Lloyd’s wanted to try to predict the winner of the 2014 World Cup by the players’ insurable values as a publicity exercise. And the CEBR was building a model to try to do that, but didn’t have access to reliable salary data, the single most important input.
No problem, I said.
Long story short: I supplied the data to the CEBR, and they fed it into the model. They then realised they wanted a lot of other metrics they didn’t readily have, about the players’ ages and contract lengths and football inflation, and I gave them that too.
The upshot: Lloyd’s were able to predict that Germany, then the fourth favourites to win the World Cup, would win the 2014 World Cup. And that prediction proved right, and Lloyd’s got lots of coverage afterwards, which was the point.
We did it all over again in 2018, and again Lloyd’s were able to predict that France, then the fourth favourites to win the World Cup, would triumph in Russia. And they did.
Together we went for a hat-trick in 2022 for the Qatar World Cup, but England and Brazil, the model’s choice of finalists, both fell in the last eight, while the model’s third and fourth favourites, France and Argentina, ended up contesting the final.
I don’t know how instrumental it ending up being, but deadline pressures in 2022 meant that the salary data for the model needed to be based upon provisional squads of 35 or so players per nation, not the final 26-man squads; in 2014 and 2018 we used the final 23-man squads.
Using provisional squads may or may not have been significant in 2022. We’ll never know. It’s difficult to make really accurate predictions for any sporting event, at any time. If it weren’t, we’d all be gambling billionaires.
So to Euro 2024. After a decade of experience working with the CEBR and Lloyd’s, Sporting Intelligence will be trying to forecast various outcomes at Euro 2024 with our own model.
By various outcomes I mean not just the winner of the tournament and the teams that will or won’t progress to the last 16 and beyond, but more granular detail about how and why each nation might succeed or fail.
This piece today is free for everyone to read. I’m going to use it to explain the rationale behind the model and the inputs that underpin it. I think it’s really simple. But having written about another analytical subject I thought was really simple not that long ago (“To xG or not xG: that is the question”), I shouldn’t take anything for granted.
From tomorrow, for six days straight days, I’ll be digging deep into the data to try to predict what might happen in each of Groups A to F at the Euros, but also exploring why it’s not as simple as saying the nation with the best squad (most valuable squad by insurable value) will win any given game.
Tomorrow, for example, I’ll detail how in Group A, Germany have by the far the most valuable squad by insurable value, at more than £1bn, or an average of more than £40m per player. None of their Group A opponents have squads much more than a third as valuable as Germany’s, by insurable value.
If Germany fielded their most valuable XI, they’d have players with an insurable value of more than £850m on the pitch, and if they field what I think is a more likely XI, it will still exceed £600m. The most valuable XIs of any of their rivals in their group are worth only around 40% of a likely Germany XI. I’ll be providing actual numbers in detail for all the above tomorrow.
You’re probably thinking by now: “What does insurable value actually mean?”
Great question.
To simplify, it’s a figure that the model expects each player to earn between now and the end of their playing career, from activities where the earnings considered will come directly from still being a professional footballer. Or in other words, playing salaries (the vast majority of most player’s value) and any endorsements directly related to still playing, for example a boot deal that remains in place only while still playing.
How does the model arrive at this figure? By considering a player’s age, and talent, and potential to have a long career, combined with a random element that they might either suddenly thrive against expectation or crash and burn unexpectedly. And future inflation, which is notoriously hard to get right. (But if it’s wrong, it will at least be wrong to precisely the same degree for every player).
How do you quantify talent, the most important metric, in combination with age? Via two related but separate metrics: wages, and the level at which each player is currently performing. Why wages? Because all other things being equal, wages are a proxy for talent. The better you are, the bigger the club you get to join, and the higher your wages rise.
Why the level at which you’re playing? If you’re a Real Madrid or Manchester City regular, and winning domestic trophies often and going deep into European tournaments, the chances are you’re a highly talented footballer. If you’re at a small club in a minor league, the chances are you’re not.
If this is already too technical, apologies. What I’m trying to convey is that a really young player, already playing at a really high level, on big wages, at a massive club, in a manner where they can demonstrably handle that environment, then their insurable value is going to be high.
To name just a few players in this category, all still to reach their 23rd birthdays, think about the likes of Warren Zaïre-Emery, Jude Bellingham, Jamal Musiala, Pedri, Eduardo Camavinga, Joško Gvardiol, Bukayo Saka and Gonçalo Ramos.
Players a tiny bit older but also worth a packet include William Saliba, Phil Foden, Dušan Vlahović and Aurélien Tchouaméni.
Every day in this Euro 2024 series, I’ll be giving you loads of detail, and numbers, about these potential stars of the show and many more besides.
At the other end of the age spectrum, there are some really good players, but already past the official retirement age of a professional footballer (35), so their insurable value is pretty much whatever their current club is going to pay them until the end of their current contract, give or take.
It’s a quirk in the model and I’ve attempted to iron out quirks - but no model is infallible and I’ll be transparent about the strengths and weaknesses. I hope you’ll enjoy the series as an alternative tournament preview.
While insurable value by the definition of Sporting Intelligence’s 2024 model is quite different from the transfer value model of a body such as the CIES Observatory, in a lot cases the outputs are quite similar. In others they are hugely divergent. That’s because the CIES gives strong consideration to contract length for all players.
Jude Bellingham, who the CIES rate as a player worth more than 250m euros today, has a Real Madrid contract to 2029. If his contract there expired in summer 2025, he’d be worth a fraction of that by their model.
His value according to the Sporting Intelligence model is … going to be revealed on Monday.
Tomorrow I’ll focus on Group A, then Group B on Sunday, Group C on Monday, Group D on Tuesday, Group E on Wednesday, Group F on Thursday and a “wall chart” overview of how the tournament will unfold on the day it starts, next Friday. Which is where we came in.