This blog is essentially for sidebars of stuff on my regular blog. If I think something is a little esoteric for the narrative of that blog, it goes here
What I'm presently reading:
The Gangs of New York by Herbert Asbury
Tea from an Empty Cup by Pat Cadigan
The Seduction of Place by Joseph Rykwert
Islands in the Net by Bruce Sterling
The Character of Cats by Stephen Budiansky
The Confusion by Neal Stephenson
Bringing Down the House by Ben Mezrich
The Zenith Angle by Bruce Sterling
The Virgin Suicides by Jeffrey Eugenides
Search this site

powered by FreeFind
Michael Jennings Extra

Tuesday, March 11, 2003
I made some comments on my regular blog about not being entirely happy with the way in which the World Cup points system chooses between sides who score an equal number of points. I think this is the situation in which the complaints about the points system being "too complicated" do have a little justice. For the record, the playing regulations actually say the following.

Okay, now let us look at this in detail. Firstly, if two sides have the same number of points, then the side with the most wins goes through. This means that if side A has four wins, and side B has three wins and 2 no results or ties, then side A goes through. Now, the question to be asked is, does this actually prove that side A is better than B. I don't think it does. If side B has had 2 no results, all this means is that side B has been unlucky to have its matches rained off. It proves nothing at all about how good side B is. It may be that side B was clearly better than its opponents and would have won both matches and gone through ahead of A, or it may be that side B was clearly worse than its opponents and would have lost both matches. However, we just do not know. All we can do is split the points. The "number of wins" test says nothing about the relative quality of side B over side A, so I think it is wrong to use it.

However, if sides have the same number of wins, we next go to the result of the head to head match between them. If A beat B, then A goes through over B. This sounds fair, but I am not sure that it is. If A beat B then for the two sides to have the same number of points there is some other side C for which B beat but A lost to. So on the direct A beat B match, we conclude that A is better than B, but on the B beat C beat A basis we conclude that B is better than A. Again, I do not think using the head to head result proves anything conclusive.

Now this is still reasonably okay if we only have two sides on the same number of points. If we have 3, then often we have two possibilies. Either A beat B, A beat C, and B beat C, in which case A goes ahead of B which goes ahead of C. What to do in this case is straightforward, but I am still not sure it is fair, because to have the same number of points, C must have the best results against other sides, and why should this not be rewarded. If (as happens more often) A beat B beat C beat A then we cannot conclude anything. Usually we go through to net run rate. If there is only one side to go through out of the three, then we simply choose the one with the best runrate. However, if there are two teams to go through, then the first team to go through is the one with the best runrate. (Let this be A). For the second team to go through, however, do you choose the team with the second best runrate, or do you go back to the head to head result between B and C. The present rules say that you go to the second best runrate. However, the disadvantage of this is that the number of points scored by A can influence whether B comes ahead of C or behind C. If A scores the same number of points, B beat C, and C has a better runrate than B but not A, then in the case of a three way tie C goes through ahead of B but if A scores more or less points against other sides, then B goes through ahead of C. I do not think the performance of A against other sides should affect the positions of B and C relative to each other. However, in the present circumstances it can. (This anolmaly was responsible for the fact that at one point it looked like it was better for England to lose their last game than win it).

If we have four or more teams on the same number of points, these problems get worse, but are essentially the same. In this tournament, this looked like happening, and Sri Lanka tried to get a ruling from the ICC that in such a case, after the first qualifier was decided, the decision should go back to head to head, but the ICC ruled against them. Fortunately, it did not happen.

Basically, I don't think that the rule of choosing the winner of head to head games is useful either. I don't think it can conclusively say that A has played better than B, so I think it should also be abandoned.

After these methods for separating sides, we go to net run rates. A higher net run rate means that a side won its matches by greater margins, or lost them by smaller margins. As far as I can see, a higher net run rate genuinely does mean that a side has played better in the tournament so far. As such, I think it is a fair way of separating sides. If net runrate is the only means used for separating teams, there are no circumstances where a side can improve its chance of qualifying by losing or winning more slowly. (There are some circumstances where sides may want to bat more slowly in order that a preferred opponent may qualify, but this is the only problem I can see).

Plus, using net runrate only has the advantage of simplicity. It is always going to be able to separate teams, and having only one rule means that you are never going to get oddities caused by the interaction between multiple rules, as now.

This is all without bonus points, which are thankfully not being used in this tournament. If we introduce bonus points, things can sometimes get worse, and sometimes better, depending on how the bonus points are awarded. I wrote some comments on this on Usenet about a year ago, and I am glad to see that one of my suggestions has since been adopted.
This page is powered by Blogger, the easy way to update your web site. In Association with

Comments system thanks to Yaccs


e-mail me  |   Home  |   Archives