17 November 2022

It's coming home, or is it? Actuaries get to grips with FIFA World Cup 2022 predictions

As the 2022 World Cup, arguably the greatest sporting show on earth, gets underway, the game of predicting the winner has also begun. Will Brazil claim their sixth title? Will England go that extra mile after a semi-final defeat in 2018 or will Wales, in their first World Cup since 1958, surprise the world? 

Actuaries, who usually spend their time dealing with the nitty gritty of risk modelling, are using their skillset for more jovial purposes to predict who will be crowned world champions on 18 December.

Prior to the Euro 2020 final, Addactis correctly predicted Italy would triumph over England. The French actuarial consultancy and technology vendor’s software simulated 10m possible game outcomes based on overall offensive and defensive efficiency of each side, and teams they played over the preceding three years.

Lloyd's of London

Based on collective insurable value of the players in each squad, Lloyd’s of London’s has predicted England will go one step further than they did 18 months ago and defeat Brazil in the final. Using this model, Lloyd’s correctly predicted Germany and France as winners in 2014 and 2018 respectively.

England’s squad boasts the highest insured value of £3.17bn ($3.78bn), beating France and Brazil’s squads with values of £2.66bn and £2.56bn respectively. Or in more blunt terms, the average insurable value of one England or France player is greater than the value of the entire Costa Rica squad.   

Lloyd’s estimates insurable value based on metrics such as salaries, sponsorships, on-field position as well as age with players between 18-24 having an average insurable value of £32m compared to £12m for players over 31.

Crowe

Risk consultancy Crowe predicted Brazil would take the title.

Lloyd Richards, head of actuarial and Arsenal fan, employed a traditional multivariate regression model using historical scores, expected goals scored and possession per game. In this model, Richards also incorporated a judgement-based approach to allow for the temperature in the host country – which northern hemisphere teams may struggle with – as well as injuries to key players.

His colleague Juweena Appanah, senior risk consultant and Liverpool fan, took to Python and machine learning technology to develop a multiple linear regression model which added number of corners, yellow and red cards to the criteria. The end result was the same though: it predicted Brazilian victory.

Other notable forecasts from Crowe’s traditional model include England, along with traditional heavyweights Netherlands and Germany, failing to get out of the group stage.

Richards told InsuranceERM the firm’s data set comprised of results from just under 2,000 previous games, comprising those at the last four World Cups as well as all 2022 World Cup qualifying fixtures. It also incorporated the results of other most recent international tournaments including Euro 2020, the 2021 Africa Cup of Nations and the 2021 Copa America.

Bookmakers and sports data firm Stats Perform have also ranked Brazil as most likely to win.

So why could Lloyd’s of London be such an outlier?

Richards reasons: “There is a premium on the value of English players because of the Premier League rules requiring a certain proportion of home-grown players and the Premier League being by far the richest league in the world.

“Inevitably an English player is worth a lot more than an equivalently talented Spanish or Brazilian player.”

World Cup 2022 tournament favourites

Bookmakers’ rankCountryCrowe’s traditional modelCrowe’s  machine learning modelLloyd’s of London insured value
Brazil  1st  1st  2nd 
Argentina  R16  R16  4th 
France  4th  3rd  QF 
England  Group  4th  1st 
Spain  QF  2nd  3rd 
Germany  Group  QF  QF 
Netherlands  Group  QF  QF 
Portugal  R16  QF  QF 

An England fan's perspective

By request of my InsuranceERM colleagues, I’m removing my journalist hat and becoming solely a football (and Arsenal) fan in making my own prediction for England’s chances.

Like Crowe’s traditional model, I previously thought England would not get out of the group stage.

However, closer inspection of our group B opponents (USA, Wales and Iran) made me think we should, hopefully, qualify from the first round.

Beyond this though, I’m dubious of our chances. Despite being runners up in Euro 2020 and semi finalists in the 2018 World Cup, I cannot see us being able to compete with teams such as Brazil and Argentina who, in my opinion, have improved significantly in the last four years.

If England were going to claim glory, the time to do so was in either, or both, of the last two major tournaments.