Qualifying results have no bearing on major tournaments

CROATIA had the second-worst record of European qualifiers for the last World Cup. They finished behind Iceland in a weak group. At the World Cup, Iceland and Croatia were drawn in the same group again. Iceland finished bottom, Croatia top. Croatia went on to reach the final.

When trying to anticipate what might happen at a World Cup it seems only natural to look at how teams got there. It is completely natural and completely wrong.

There is next to no correlation between how a team perform at a World Cup and how they performed beforehand.

I analysed the last six World Cups – qualifiers and tournament.

Those were the World Cups from 1998 to 2018. At each tournament there were 32 teams, with either 13 or 14 qualifiers from Europe.

I ranked European qualifiers for each tournament by points per game in their qualifying group. The team with the highest figure I ranked first, the team with the lowest figure I ranked last – teams with the same figure I separated on goal difference per game.

After each World Cup, competition organisers Fifa rank teams at the tournament from 1 to 32 – first for the winners, second for the runners-up, and so on all the way down to 32nd for the team knocked out in the group stage with the worst record. I used those rankings to put European teams in order.

So I had ranks for European teams in qualifiers and ranks for European teams at the tournament. There was next to no correspondence between them.

The relationship between two sets of numbers can be measured mathematically by something called a correlation coefficient.

If results in qualifiers had told us everything we needed to know about what would happen at the tournament, the correlation coefficient would have been 1.0 – the highest ranked qualifier would have been highest ranked at the tournament, and so on.

If results in qualifiers had told us nothing about what would happen at the tournament, the correlation coefficient would have been 0.0 – the lowest ranked qualifier would have been as likely as the highest ranked qualifier to be highest ranked at the tournament.

What was the correlation coefficient? It was less than 0.1. Let me give you one example of the about-turns that made this number so low: qualifiers who had ranked in the bottom four reached the semi-finals or beyond more often than qualifiers who had ranked in the top four.

France ranked 12th of 13 among European qualifiers for the 2006 World Cup but second at the tournament. Germany ranked 11th of 14 among European qualifiers for the 2002 World Cup but first at the tournament.

Germany were beaten in the 2002 final by Brazil, who had finished third in South American qualifying. There are fewer countries in South America than in Europe and it has one qualifying group for them all. In 2010 Argentina qualified in fourth place, Uruguay in fifth and last place. At the tournament, Argentina were the second-best South American performers and Uruguay the best.

What is the relationship between results in qualification and results at a tournament? For all practical purposes, there is no relationship whatsoever.