Soccer’s first Winter World Cup is fast approaching. With less than two weeks to go until Qatar and Ecuador get the tournament underway, it’s time to put our supercomputer1 to work and start predicting how things will play out. Who will make it all the way to the final and which big names might fall short? Here’s a primer on how the world’s greatest sporting spectacle may unfold based on historical data.
Check out our interactive visualization and explore every team’s chances of advancing past the group stage and eventually winning the tournament. We will update the probabilities after each game. We’ve also written a blog post in Dutch which focuses on Belgium.
A World Cup often shapes up as a battle between Europe and South America. Historically, teams from either continent pretty much used to alternate wins. However, by the time the 2022 edition kicks off in Qatar, it will have been over two decades since Ronaldo’s goals propelled Brazil to their latest World Cup victory in 2002. Over the next four editions, only European teams brought the title home, with a peak in 2018 when all four semi-final spots were filled with teams from the old continent (France, Croatia, Belgium and England).
Who is most likely to win this World Cup?
In Qatar, the tables might turn again. According to our simulations, Brazil is the tournament’s main favorite with a 21% chance of winning. The Seleçāo certainly look equipped to break the European dominance of the World Cup. Coach Tite has the Brazil attack firing on all cylinders. They scored 40 goals in 17 qualifying games — 13 more than Argentina, the next-most-prolific CONMEBOL side. Their attacking options of Gabriel Jesus, Richarlison, Vinicius Junior, Roberto Firmino, Raphinha, Rodrygo, Antony and star player Neymar offer a multitude of offensive options and provide goalscoring threat from every kind of situation. Moreover, Brazil rarely concedes a goal. Our model rates their defense led by veterans Thiago Silva and Dani Alves as the most solid one across all World Cup teams.
| Chance of reaching round | ||||||
|---|---|---|---|---|---|---|
| # | Team | 1/16 | 1/8 | 1/4 | Final | Win ▼ |
| 1 |  | 93% | 71% | 46% | 32% | 21% |
| 2 |  | 91% | 65% | 39% | 26% | 17% |
| 3 |  | 91% | 59% | 39% | 20% | 11% |
| 4 |  | 89% | 57% | 36% | 18% | 9% |
| 5 |  | 83% | 49% | 27% | 15% | 7% |
| 6 |  | 81% | 54% | 29% | 14% | 7% |
| 7 |  | 76% | 39% | 21% | 12% | 6% |
| 8 |  | 84% | 53% | 27% | 12% | 5% |
| 9 |  | 76% | 43% | 21% | 11% | 5% |
| 10 |  | 81% | 44% | 24% | 10% | 4% |
Argentina is the only team that has beaten them over the past four years in a non-friendly game — in the 2021 Copa America final. That makes Argentina a major contender with an 11% chance of winning. With Lionel Messi, they have an all-time great leading their squad. In what will probably be his last tournament, the superstar will be eager to go out on a high. The final could even be the tailpiece of the Messi vs Ronaldo rivalry. Just like Messi, Ronaldo will most probably play his last World Cup, considering the age of the 37-year-old star. Therefore, Argentina will first have to get past Brazil, though. In the most likely scenario where both South American teams win their group, they will meet each other in the semi-finals.
Spain enters the tournament as the highest-rated European side with a 17% chance of winning. Indeed, Spain has plenty of talented players, among them the Barcelona midfield duo Pedri and Gavi. However, La Roja will have to get past Brazil and Argentina to reach the final. Hence, it risks premature elimination.
Defending champions France has the fourth highest odds (9%) and a comparatively easier path to the final compared to its top-notch rivals. According to our model, they need to eliminate England and Portugal in the quarter and semi-finals respectively and neither of those teams is in top form. However, it was recently announced that France will be without the contribution of their star midfielders Paul Pogba and N’Golo Kanté due to injury, which are critical absences, particularly remembering their spectacular performance in Euro 2020.
While Belgium was among the favorites in 2018 and the delayed Euro 2020, they check in with the seventh best chance in 2022 as several recent losses have seen their ratings decline. Kevin De Bruyne remains in top form having finished 3rd in the Ballon D’Or, but their core is aging and there will be worries about Eden Hazard’s lack of game time and Romelu Lukaku’s recent injury setback. This tournament may be the last one that their golden generation plays together.
After the top four contenders, European teams (Portugal, England, Belgium, Netherlands, Germany, and Denmark) round out the top ten with chances ranging from 7% (Portugal, England) to 4% (Denmark). No team outside of South America or Europe has a >1% chance of winning.
How our predictions work
At the core of our forecasts are three rating systems that estimate a team’s strength:
- First, we compute an Elo rating from recent international match results. These results are weighted for the recentness of the game, the importance of the competition and the strength of the opponent. We are more aggressive in our updates than most traditional Elo rating systems. Therefore, our ratings reflect a team’s recent form, rather than its longer-term evolution.
- Second, we assign individual offensive and defensive ratings to each team based on a team’s goals scored and conceded in recent games. Although offense and defense aren’t easy to separate in a fluid sport such as soccer, there are some useful reasons to handle things in this way. First, an important factor in soccer games is the playing style of both teams and the balance between offense and defense. This difference in playing style might be an important factor in deciding the game’s final outcome. For example, a game between two teams known to rely on a very strong defense might have a higher probability of ending up in a draw than a game between two teams that are known to play very offensively. Second, defensive ratings tend to have a little more predictive power than offensive ratings in games against elite competition, which will be the case for most key matches that will be played at the World Cup.
- Finally, we include the cumulative market value of each nation’s starting XI as estimated by the website Transfermarkt. Because most teams still have to make their final list of players official, we relied on recent matches and some of our soccer knowledge to come up with the most likely lineups for these teams. We preprocessed the data to overcome issues such as market inflation or missing values.
Below we show these ratings for each World Cup team. Because all of these rating systems are on different scales, we’ve standardized them to make them comparable to one another. Positive scores are associated with above-average teams according to the rating system in question. For instance, a score of +1 represents a team that’s one standard deviation better than the mean. Negative scores are associated with below-average teams.
| Team ratings | |||||
|---|---|---|---|---|---|
| # | Team | ELO | Off | Def | MV |
| 1 | | 2.14 | 1.63 | 2.12 | 1.65 |
| 2 | | 2.93 | 2.05 | 1.25 | 1.22 |
| 3 | | 1.48 | 0.93 | 1.50 | 1.03 |
| 4 | | 0.93 | 0.79 | 0.70 | 1.98 |
| 5 | | 0.99 | 0.78 | 0.90 | 1.80 |
| 6 | | -0.28 | 0.45 | 1.19 | 2.29 |
| 7 | | 0.53 | 2.19 | 0.66 | 0.60 |
| 8 | | 0.84 | 1.37 | 0.43 | 0.15 |
| 9 | | 0.60 | 1.26 | -0.17 | 1.54 |
| 10 | | 0.11 | 0.59 | 0.85 | 0.05 |
| 11 |  | 0.20 | -0.43 | 0.57 | 0.16 |
| 12 |  | -0.20 | 0.35 | 0.27 | -0.42 |
| 13 |  | 0.17 | 0.11 | 0.44 | -0.21 |
| 14 |  | -0.64 | 0.70 | 0.01 | 0.07 |
| 15 |  | -1.15 | -0.78 | 0.46 | -0.11 |
| 16 |  | 0.01 | -0.75 | -0.19 | -0.30 |
| 17 |  | -0.19 | -0.51 | 0.22 | -0.87 |
| 18 |  | -0.26 | -0.85 | 0.79 | -0.84 |
| 19 |  | -0.25 | -0.98 | 0.40 | -0.40 |
| 20 |  | -1.15 | 0.15 | -1.29 | -0.25 |
| 21 |  | -0.78 | 0.30 | -2.11 | 0.21 |
| 22 |  | 0.09 | -0.25 | -1.59 | -0.64 |
| 23 |  | -1.21 | -0.38 | -0.22 | -0.70 |
| 24 |  | 0.37 | -0.85 | -0.56 | -0.35 |
| 25 |  | 0.32 | -0.40 | 0.08 | -0.93 |
| 26 |  | -1.47 | -0.81 | -0.65 | -0.77 |
| 27 |  | -0.21 | -0.92 | -1.05 | -1.07 |
| 28 |  | -1.51 | -1.11 | -0.17 | -0.96 |
| 29 |  | -1.34 | -1.18 | -0.95 | -0.65 |
| 30 |  | -0.60 | -1.19 | -1.36 | -1.06 |
| 31 |  | -0.28 | -1.06 | -1.75 | -1.09 |
| 32 |  | -0.17 | -1.18 | -0.76 | -1.12 |
Usually the three ratings are strongly correlated with one another. But there are some exceptions. England, for example, can line-up the most expensive starting XI, but their Elo ranking has fallen to below average ahead of the tournament as they failed to win their last six games. Hopefully, they can revive at the World Cup, but such form is not a sign of tournament success if we go by past winners.
Given the ratings of two teams, we then use a diagonal inflated bivariate Poisson regression model to estimate the distribution of possible outcomes between two teams. The fancy model is necessary because of the complex set of tiebreakers used to determine which teams will advance to the knockout stage in case of equal points. Crucially, goals scored and goals allowed are used, so knowing the chance of a 1-0 win compared to a 2-0 win is important.
We calculate the number of goals that we expect each team to score during the match.
How injuries might affect the World Cup
Injuries or suspensions to key players can wreck a team, especially in the knock-out phase of a tournament. As an example, Belgium was our forecasted favorite for the 2020 EUROs but they suffered a loss against Italy in the quarter-finals after injuries of Kevin De Bruyne (he still played at half strength with a torn ankle ligament) and Eden Hazard — two of their key players. Admittedly, one of our model’s biggest weaknesses was that it didn’t account for squad depth, which could be especially important this time around. Players will be fatigued not only by the World Cup schedule but also by the fact that domestic schedules had to be crunched to accommodate a World Cup in Nov-Dec.
Using the lessons learned from the previous forecasting attempts, we’ve now added the incidence of injuries and suspensions as an additional stochastic event to our simulation. In each game, we knock out a random starting player with a probability of 20%2 and replace him with the highest-valued player that plays in the same position.
Interestingly, this approach makes it possible to analyze the effect of individual players on our predictions. Specifically, we’ve looked at four scenarios:
- Lukaku misses the tournament (▾ 0.20%). Belgium’s Romelu Lukaku is at risk of missing the World Cup struggling with a hamstring injury. While Belgium’s head coach recently called him “irreplaceable”, his most likely substitute would be Michy Batshuayi, who currently plays as a forward for Fenerbahce. This would decrease Belgium’s probability of winning from 5.6% to 5.4%.
- De Bruyne misses the knockout phase (▾ 0.28%). Talking about irreplaceable Belgians, one cannot go around Kevin De Bruyne. The Premier League’s most creative midfielder is a crucial element in Belgium’s team. In a scenario where he would get injured after the group phase, Belgium’s probability drops by 0.28 percentage points.
- Mbappé misses the knockout phase (▾ 0.37%). No player has a bigger impact on a team’s win probability than Killian Mbappé. Even a team with the squad depth of France cannot replace the fast winger. With a market value of €160m, he is the most expensive player at the World Cup.
- Vinicius Junior replaces Neymar after the group stage (▴ 0.30%). Interestingly, injuries of key players do not necessarily have a negative impact on a team’s odds. For example, Neymar will probably get the preference over Vinicius Junior on Brazil’s left wing thanks to his status. Yet, Vinicius Junior (€120m) currently has a much higher market value than Neymar (€75m).
Generally speaking, we found that France has the best squad depth, while Mexico and Australia have the worst. To determine this, we used a simulation-based approach. For each team, we simulated 10,000 matches where the team plays against itself. We fix one lineup to be the same in each match: the team’s projected starting XI. For the opposing lineup, we uniformly at random replace a single player in the lineup. Finally, the squad depth index is computed by dividing the difference between the number of games won by the projected starting XI and the modified lineup by the number of simulated matches.
Which teams could surprise?
Looking at each team’s change of Elo ratings since the 2018 World Cup, Canada might be the tournament’s dark horse. They have an emerging golden generation that is spearheaded by Alphonso Davies (aged 21), Jonathan David (22) and Tajon Buchanan (23) — the best young player at the 2021 CONCACAF Gold Cup.
They will compete in Group F with Belgium, Morocco, and Croatia, and there is a chance that they may be able to pull off an unexpected victory and go to the next round. In particular, their speed and adeptness at counter attacking may prove to be a good matchup against Belgium’s higher defensive line and aging defense.
Japan is also on an upward trajectory. During the last international break, they thoroughly outplayed the US and drew with Ecuador. They have a number of promising young players such as Takehiro Tomiyaus (Arsenal) and Takefusa Kubo (Real Sociedad).
The evolution of Elo ratings also highlights that the week of 16 November 2020 was truly exceptional in international football. On Tuesday, Spain won 6-0 against Germany in the Nations League, giving their Elo rating a huge boost of 471 points. On the same day, Ecuador beats Colombia 6-1 and, the day after, Serbia beats Russia 5-0.
