Just a quick note regarding the results of the World Cup. My previous post talked about the probabilities of various scenarios happening. Here’s what really happened.
The chances of a seeded team being eliminated in the group phase was estimated at 25%. And in reality, it didn’t happen. Bottom line: shock exits, which were fairly common in the 32-team format, are unlikely in the new format.
The chances of two seeded teams meeting in the round of 32 was estimated at 63%. It didn’t actually happen, but 63% is far from a sure thing. Also, note this quote from my post “You wouldn’t be out of line in predicting Spain vs. Argentina as the Final, and yet it could happen in the first round of knockouts!”. Turn out the first half of that sentence was spot on. What a different tournament it would have been if those two teams had met in the round of 32. The final probably wouldn’t have been such a dud (likely Spain vs. England?), but then again we wouldn’t have gotten the amazing Argentina knockout games versus Cape Verde, Egypt, and England.
Going into the final group matches, how many teams had clinched advancement, or been eliminated, or were still in the hunt? Here’s the table from my previous post, with the results of 10,000 simulations compared to the reality of 2026:
| Scenario | Simulated | Reality |
| Clinched 1st place | 2.2 | 4 |
| Clinched at least second place | 8.6 | 6 |
| Clinched at least 3rd place advancement | 2.1 | 0 |
| Neither clinched nor eliminated (still in the hunt) | 33.3 | 33 |
| Eliminated | 1.8 | 5 |
So reality was a little tougher on the minnows: five teams were eliminated after the first two group games, versus the sim average of 1.8. Likewise, three fewer teams clinched advancement than expected. But these variances are well within the expected levels.
In some ways, this was a very deterministic World Cup. It went basically as expected. No seeded teams were knocked out in the group phase. There were few upsets in the knockouts. The semifinals consisted of the four best teams (either by FIFA rating, ELO rating, or oddsmakers), something that had never happened before. Is this because of the 48-team format? With n=11i.e.., only one data point, since this is the first tournament using this format, who can say? Maybe after a few more 48-team tournaments we’ll have a better idea. But, I suspect a 64-team tournament may be happening sooner rather than later…
