At the end of my last article, I left with a question I couldn’t stop thinking about.
Watching this World Cup, it felt like the gap between teams was getting smaller. Cape Verde pushed Argentina to the edge. Uruguay crashed out of a group many people expected them to cruise through. France looked like a completely different team by the end of the group stage than they did at the beginning. Every few days I found myself saying the same thing: these games just feel closer than they used to.
Of course, watching soccer and proving something are two different things. So I went back to the model.
Instead of training it on previous World Cups, I retrained the exact same regression using only the matches we had just watched in 2026. I didn’t add any new variables or try to make it more sophisticated. I wanted to know whether the relationship between FIFA rankings and goal difference looked any different after this tournament.
I honestly expected it to get stronger.
These rankings had just been tested in real matches. If they were doing a good job describing international soccer, I thought they would explain this World Cup at least as well as they explained the previous ones.
They didn’t. The R² dropped from about 0.19 to roughly 0.10.
In practical terms, FIFA rankings explained about half as much of the variation in goal difference during this tournament as they did in the historical World Cups I originally trained on.
The overall trend was still there. Higher-ranked teams generally outscored lower-ranked teams. What changed was how often the actual results drifted away from what the rankings suggested.
Germany’s 7-1 win over Curaçao was much larger than the model expected. Canada’s 6-0 win over Qatar was another example. On the other end were matches like Spain drawing Cape Verde 0-0 or Iran dropping points to New Zealand. Looking at all of those games together, the tournament simply had more variation than I expected.
Once I saw the results, I wasn’t actually that surprised anymore. They matched what I had spent the past few weeks watching.
Before the tournament, I assumed the new 48-team format would make matches easier to predict. If the fifth-ranked team in the world is playing someone ranked in the 70s, that should be more straightforward than two teams sitting next to each other in the rankings.
Now I’m not so sure.
Larger ranking gaps seem to create more extreme outcomes in both directions. Sometimes the favorite wins by six goals. Sometimes they spend 90 minutes trying to break down a disciplined defense and walk away with a draw. Rankings still tell us which outcome is more likely, but this tournament reminded me how much room there is for things to go differently.
Cape Verde kept bringing me back to that idea.
When I looked back at their tournament, I realized they hadn’t exposed some fatal flaw in FIFA rankings. Before the tournament started, every piece of information we had suggested they were the weakest team in that group. That wasn’t an unreasonable conclusion.
Then the games started. Over three matches, Cape Verde gave us completely new information. They defended brilliantly, frustrated better teams, and showed they were much stronger than anyone expected.
That ended up changing how I think about prediction models. The goal of a prediction model is to predict the future. But it can only make predictions using the information that exists before kickoff. Before this tournament, there wasn’t any evidence suggesting Cape Verde would frustrate Spain, eliminate Uruguay, and push Argentina to the final whistle. Those performances became new information. Every World Cup teaches us something we didn’t know before it started, and that’s why I think retraining models after a tournament is almost as interesting as building them beforehand.
France made me think about this in a different way. Before the tournament, I leaned too heavily on what I’d seen from them at Euro 2024. They looked slow, they weren’t creating many chances, and I questioned whether they deserved to be considered one of the favorites.
Looking back, I think I focused too much on one tournament instead of the bigger picture.
France looked better every game they played. Against Senegal, I remember thinking they looked surprisingly poor in the first half. I still had an inkling they would grow into the game, so at halftime I actually put money on France. They did exactly that. By the end of the group stage they looked like one of the strongest teams left.
That also made me realize something else. International soccer isn’t static. Teams change over three or four weeks. Some find confidence. Some build chemistry. Others never quite click.
Club teams spend almost the entire year together. National teams get a handful of international breaks and then a few weeks at a major tournament. You can’t just take the quality of each individual player, add it all together, and assume the team with the most talent will play the best soccer. France felt like a team that was figuring itself out as the tournament went on.
I haven’t seen that same progression from every favorite.
Argentina are fascinating because they’ve done everything that’s been asked of them, but I still don’t feel like we’ve learned exactly how good they are. If you look at the average FIFA ranking of the opponents they’ve faced in recent times, they haven’t the toughest opponents. Cape Verde was really the first match where they looked uncomfortable. One of the commentators said after the final whistle that Argentina survived more than they won, and I thought that was exactly right. They’re absolutely capable of winning the World Cup, but I’m much more interested in seeing how they look once they’re facing teams that can consistently challenge them.
England leave me with a similar feeling. They’re still alive, but they haven’t looked convincing. At some point they’re going to need more than a Harry Kane moment to bail them out, like they did against DR Congo. The deeper you go into a World Cup, the fewer opportunities you get to scrape by.
Portugal are probably the team I’m most conflicted about. On paper, they may have the deepest squad left in the tournament. Everywhere you look there’s another world-class player. But when I watch them, I don’t see the same chemistry I see with France. Against DR Congo and again against Croatia, it felt like a collection of great players more than a great team. Finishing second in the group made their route through the knockouts much harder than it needed to be, and I can’t help wondering whether that’s more about chemistry than talent.
That might be one of the biggest limitations of FIFA rankings.
Rankings do a good job measuring what a national team has accomplished over time. They don’t tell us how that team is changing during a tournament.
When I started this project, I thought I was trying to answer a question about FIFA rankings. Looking back, I think I was really trying to understand what makes international soccer so difficult to predict.
The rankings still matter. If I were building another model tomorrow, they’d be one of the first variables I’d include. After I retrained this one, my brother and I spent an evening at a whiteboard thinking about what else we could feed into it: recent form, club performance, market value, injuries. Each variable sounded promising for about five minutes, until we found a reason it couldn’t fully explain what happens at a World Cup.
I think that’s because international soccer is asking a different question than club soccer.
At the club level, the most talented teams usually have months, sometimes years, to build chemistry and establish an identity. National teams don’t have that luxury. They get a handful of international breaks, and then a few weeks together before they’re expected to compete with the best in the world. That’s why you can’t just add up the talent on a roster and assume the most gifted team will play the best soccer.
Soccer is a team sport in the truest sense of the phrase.
France often looks like a team that grows into itself over the course of a tournament. Portugal still feels like it’s searching for that connection. Cape Verde doesn’t have the same level of individual talent, but they play with a level of organization and belief that can make far more recognizable teams uncomfortable.
I watch Premier League soccer almost every weekend, and I still think the quality of play at the club level is higher than what we see internationally. But there’s something about players representing their country that’s hard to quantify. Maybe it’s pressure. Maybe it’s emotion. Maybe it’s the fact that they only get a handful of chances to do it.
Whatever it is, players seem to find another level. And that’s why every World Cup gives us a Cape Verde.
It’s also why—I think—we keep coming back every four years.
