Group Stage Scoring Models

The following is partly obsolete thanks to a late change in the Regulations by FIFA. More news in our Blog Page.

For 2006, FIFA decided that teams tied by points will be settled by a head to head system: "If two or more teams are equal on the basis of the above criterion," [equal in points obtained in ALL the group matches] "their ranking shall be determined as follows: (a) greater number of points obtained in group matches between the teams concerned; (b)...."

It's clear that the head-to-head comparison applies only to teams equal in points in the overall group — i.e., only one mini-group is formed of two, three or four tied teams unless there are two two-way-ties (two tied for first and two tied for third place), in which case it becomes two parallel but separate mini-groups. From there it goes to Goal Difference then Goals in the mini-group followed by the same in the overall group and drawing of lots.

To check 'what might have happened' under different Regulations [pdf], the Ultimate Couch Potato spreadsheet allows you to choose between three models by entering 1, 3, or the default 2 in the model cell at the top of the GroupResults page, as follows:

Model Differences.

For the most part, the models should yield similar results barring the odd flip-flop in a tie. The Overall Group Model is simple to work out but gives an edge to the strongest team if they score big against the weakest. A head to head system remedies that at the expense of simplicity. Once you've dispensed with simplicity, however, you might as well go all the way and allow mini-groups to cascade and settle further ties.

In a 4-way tie for example, the mini-group is the same overall group so the latter's Goal Difference and Goal counts apply immediately. But having then used the one mini-group to apply GD and G, any further ties including a possible 3-way tie go straight to drawing lots!

Christopher Bird calculates that the 2006 model increases the likelihood of drawing lots. Considering the expense, effort and interest in the World Cup, it would be a pity if a qualifier earned its place by 'lots' instead of some performance-related measure or event such as a penalty shootout. The interpretability of the Regulations particularly in a 4-way tie swayed me to include different models in the spreadsheet. They add some flexibility to the outcome in view of the inflexible 'lots' algorithm below.

Drawing Lots

If and when lots are required to settle a tie, the spreadsheet will request a precise solution. For example, in a 1-2-2-2 [3-way tie] scenario, all it needs to know is who qualified in second place and it will demote the other 'seconds' to a third place tie. There's no reason to draw lots for third and fourth places as they are deemed simply to have not qualified. In the example above, anything but a '2' entered in a tied row is redundant. Which also means ... there's no way to make an untied third place finish into a first ... so if the model is wrong, ç'est fini. There is no remedy.