Attn: UPSF

With the success of the UP baseball team, I've been paying much more attention to baseball RPI sites than I ever have before. I came across a site called Boyd's World that has some really interesting things to say about the problem of using RPI as as model and his creation of an algorithm that helps eliminate those problems:

http://www.boydsworld.com/baseball/faq.html

Anyway, seemed like something right up your alley and figured I'd pass it along.

Interesting rating system. This sentence stumped me, however:

One of the basic tenets of the ISR's is that each game is worth the same amount.

Well, no. The games later in the season between two good teams are worth more that if the teams played each other when everyone is even at the beginning of the season. Also the games early in the season between a really good team and a stinker is worth the same as a game between two really good teams.

Clearly, in this system if everyone starts even, you want to play the bad teams early, and the good teams late, before their rating value is subject to wins and losses. Unless you readjust the early games according to season results, I don't see how it could work.

I think what he meant by that was that pollsters tend to look past mid-week baseball games while focusing on weekend series, mainly because the weekend series feature the best pitching. This formula treats mid-week games the same as weekend games. I think anyway.

To me, the thing that piqued my interest the most was what he said about trying to properly evaluate teams in baseball that play such regional schedules, a problem also faced by women's soccer. Would be interested to learn more about how he does that.

There are a lot of comparisons between baseball and soccer that are hard to equate.

First off, baseball plays 50-56 games, which should make an RPI more accurate. but I have no sense of how much teams travel across country. My guess is, not much. UP plays about 8 games in Utah, and the rest is West Coast. But, many of those games are part of 3 game weekends (we played 4 games with Seattle), which may decrease the upset probability for statistical purposes, but don't really add to the comparisons between teams in common. But we add nothing this year to comparisons to Eastern teams.

Women's soccer is even worse. we only plays 20 season games , and of that, only 2-5 games are usually out of region. many schools play almost no games out of region. UP travels more than most, and we really only get about 4 games a year with Eastern teams of teams from the middle states.

Next year's 7 games with out of region teams looks to be more than the vast majority play, and maybe more than we have played in a while. Some years we only get around 4 teams.

Any way you look at it, it's a teeney sample. No matter how you stir the data on 20 games, you have a high statistical error when you are trying to rank 320 teams, especially when half or more are conference games.