Predicting the Women’s World Cup

The plan

During the 2018 World Cup I was messing about with a simple prediction model based on the Elo ratings and came up with graphics like this and this which people quite liked, so I thought I’d try it again for the Women’s World Cup this year.

What I’ll be aiming to do throughout the tournament is create updated graphics after every round of matches, showing the revised prospects for the teams who just played and also how the latest results have impacted the knockout stages.

The model

Fortunately for me, FIFA have since adopted an Elo-based system for their rankings, so I’ve taken their latest published ratings as the basis for my model. They’re a few months out of date, but most teams will only have played a few friendlies since then. I’ve added 100 points to France’s rating to reflect home advantage, which is how the Elo ratings site recommends handling it.

As I’m not a statistician by trade, the model almost certainly isn’t as fancy as e.g. the FiveThirtyEight one, but it looks to be giving pretty similar results overall. I also took a look at the Gracenote model and this nice thread by Luke Benz (who’s done something similar to me) and there don’t appear to be any massive disagreements, which is reassuring.

How the knockout stages could pan out

There are a few fairly fundamental differences to the format of the men’s tournament that I spotted when I was building this and thought were worth flagging:

  1. As this is a 24-team tournament, the four best third-placed teams also qualify from the group stages to fill out the Round of 16. The combination of third-placed teams that qualify affects which match they each go through to, as per this reference chart.
  2. A knock-on effect of this – and having six groups feeding eight knockout ties – is that there are more possible paths for each team. Therefore you’ll see the same nation pop up in a few different places on the graphics below, including on different sides of the same fixture.

Without further ado, here’s the probabilities look for the latter stages of the tournament. France’s path looks the least uncertain: assuming they win Group A and dispatch the third-placed team they’d face in the Round of 16, they’ll probably have to get past the USA in the quarter-finals and England in the semis, with Germany their most probable opponents in the final.

… and here’s how the Round of 16 fixtures are likely to be populated. The horizontal lines in the middle pair the matches up by which quarter-final game they feed into.

Team-by-team graphics

Below I’ve created a graphic for each team (in alphabetical order) breaking out their prospects and likely opponents at each stage in detail. One slight quirk to be aware of is that the spread of possible opponents in the final for the lowest-ranked teams may be a bit off, as there are so few scenarios in which they get this far.