SRI Rankings Methodology

SRI = Spectral Radius at Infinity. Spectral radius algorithms have had much success with college sports and some other sports in giving a mathematically reliable ranking of the teams/players involved.

For a relevant reference, see "The Perron-Frobenius Theorem and the Ranking of Football Teams" by James P. Keener, SIAM Review, v.35 no.1, p.80 (1993) (please note that I have been unable to locate a copy of this paper online)

The SRI system is a rigourously mathematical way of handling results in most sports. The paper I reference above is a good start, but I've also developed a way of valuing strong wins higher than weak wins in a statistically sound (I hope) fashion. For example, in tennis a 6-0 6-0 win is worth the maximum 1 point, while a tough 6-7 6-3 6-4 result will get you approximately 0.67 and your opponent 0.33. These numbers are based on the probability that the result could be generated by two players of equal strength. The more one-sided the result, the less likely it is to be due to chance.

The SRI procedure takes into account strength-of-schedule in an optimal way. A matrix of teams' (players') results against every other team (player) is calculated, then this matrix is taken to an infinite power (with some normalisation). The results are often uncannily accurate.

All matches can be given equal weight, or time-weighted, or whatever. The final ranking is reasonably self-consistent, in that a good result against a weak team (player) is not as valuable as a good result against a good team (player), but it is the algorithm which determines who are the good and poor teams (players) by their results against everyone else.