2018 World Cup preview 2 – Projected outcomes

Here’s my full projection for the 2018 FIFA World Cup

world cup projection

This is a result of 5000 simulations, using my International ratings. I’ve put some notes about the methodology behind these below.

International football is difficult to analyse. Unlike domestic league football, where vast quantities of data exist to assess team and player performance in a variety of conditions, negligible relevant data is available for international football. It is, by its nature, infrequent – so tournament line-ups may bear little resemblance to those of previous matches – with often little or no data to assess how teams fare in truly competitive circumstances.

Some assessments are readily available online – e.g. FIFA and ELO ratings. But these aren’t perfect (particularly FIFA) and for my projections I need ratings in terms of attack and defence strength. So I’ve created ratings from scratch, using data that’s easily available.

My ratings are based on the following:

1. The attacking and defensive strength of the club side each player represents.
This is calculated using average goals for and against, for that team in the previous season. However, clearly a player in Spain’s La Liga will likely be paying at a much higher level than someone in Guatemala’s Liga Nacional. So I’ve applied factors to account for the relative strength of each league – with La Liga the highest followed by the English Premier League and German Bundesliga. I’ve tried to make this as objective as possible, using information such as UEFA coefficients and other assessments of relative football league quality – but there is inevitably some subjectivity.

2. The age profile of each team – with 26 considered the prime age, with factors decaying at ages above 30.

3. The international goal-scoring record of each player.
I’ve used this as a measure of goal scoring effectiveness at international level, weighted by the number of international games played.

The overall team ratings are an average for the best goalkeeper, midfielders and forwards.

This produces following ratings:

Country GF GA GD
Brazil 1.99 0.78 1.21
Spain 2.02 0.91 1.11
Germany 1.98 0.99 0.99
France 1.98 1.03 0.94
Belgium 1.82 1.01 0.81
Argentina 1.75 0.99 0.76
England 1.67 0.94 0.73
Croatia 1.58 1.02 0.56
Colombia 1.60 1.17 0.43
Portugal 1.48 1.07 0.41
Uruguay 1.36 0.98 0.39
Switzerland 1.38 1.42 -0.04
Serbia 1.17 1.22 -0.05
Mexico 1.43 1.50 -0.06
Poland 1.43 1.53 -0.10
Morocco 1.19 1.36 -0.17
Denmark 1.34 1.54 -0.20
Egypt 1.10 1.32 -0.22
Russia 0.96 1.21 -0.24
Senegal 1.28 1.56 -0.27
South Korea 1.03 1.42 -0.39
Iran 1.08 1.48 -0.40
Japan 1.17 1.61 -0.44
Tunisia 1.00 1.46 -0.46
Saudi Arabia 1.06 1.56 -0.50
Nigeria 1.13 1.65 -0.52
Sweden 0.95 1.49 -0.54
Costa Rica 1.02 1.62 -0.60
Iceland 0.95 1.56 -0.61
Peru 1.00 1.67 -0.67
Australia 1.06 1.84 -0.78
Panama 0.90 1.97 -1.07

These ratings aren’t perfect. One obvious omission is that they take no account of recent team form or tactics, the team ratings are based solely on the quality of the underlying players. But I think they’re a good starting point for assessing the likelihood of different world cup outcomes (and potential betting value).

I used a similar method for Euro 2016, which worked reasonably well – for example identifying value in Portugal (but also Spain).

Advertisements

One Comment Add yours

  1. I have made a collection of predictions on my site and included this prediction here: https://bstat.de/doku.php/wm18/pgwm18

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s