05-22-2017, 12:53 PM
(05-22-2017, 11:20 AM)ColoradoTree link Wrote:[quote author=oldalum link=topic=16887.msg193570#msg193570 date=1495469418]I disagree with this part of your post though. As I've commented before, a ranking system that consistently produces clearly wrong results should be fixed or replaced. The problem is that the computer system gives you points for your "x" best wins (X increases through the season), which assumes all teams play schedules of comparable strength. That I suspect is taken from the pro tour rankings, where you get Y points for your performance in each tournament. But college players are not pros and the assumption doesn't hold. The college basketball computer ranking systems do not seem to be afflicted by this problem despite quite different strengths of schedule. I don't understand why the NCAA doesn't switch to a better ranking system that would not require fixing each year by a committee that is too risk-averse to make any significant changes.
[quote author=ColoradoTree link=topic=16887.msg193565#msg193565 date=1495466777]
The rankings system isn't too impressed by beating those teams, so not playing Indoors hurts the team's ranking. Totally reasonable that it affects the ranking
[/quote]
This is an interesting conversation that gets to the purpose of the rankings--whether it is to measure which teams have had the best years, or whether it is to determine which are the best teams. When I said it was "totally reasonable" for the ITA rankings system to give the benefit of the doubt to teams that have played better teams (and gotten more wins against good teams) over the course of the season, I was thinking of the former approach. I think UNC had a better year than Stanford did. UNC played more ranked teams and had a more impressive resume than Stanford did. Not much better, but somewhat better, and the ranking differential (UNC was 4th, Stanford 6th) seems to 59er that out (the seeding differential, #2 UNC vs. #7 Stanford, however, seemed a bit stark for such an even matchup). But Stanford just proved that it's a better team than UNC is (or, at least, it was a better team yesterday). Not much better, again, but somewhat better. After all, four of the six singles matches went into third sets.
I'm not sure how much credit the ITA should build into the rankings system to account for Stanford's postseason dominance. I'm comfortable with an approach where the ITA's rankings seek to determine who has had the best season to date, while the NCAA's seedings are the NCAA's best guess at which teams are actually the best. Of course, that's not what we have, as you noted. But it seems like the appropriate division of labor to do it that way. (I think basketball does something like this, no? If a key player is injured, it won't affect the team's ranking--i.e., whether that team has earned the right to play in the tournament based on the entirety of the season. But it will affect that team's seeding.)
Ultimately, I don't think this year is a particularly egregious example of it. There were several teams that could plausibly win the title, and Stanford is one of those teams. Whether Stanford played UNC as a slight favorite or a slight underdog doesn't mean a whole lot. Last year was more striking, when Stanford was the 15 seed. There were some complications, there, however, as Stanford not only suffered its usual rankings ding for not having as many matches against top-ranked foes, but the team also lost some more matches because Carol Zhao was off with the Canadian national team, so everyone was playing a slot higher than usual. The team was 14-5 entering NCAAs, which is a lot different than 21-2 this year. (Note: one useful change was the Pac-12's decision to change the conference tournament from a singles/doubles tournament to a team tournament. This gave Stanford a chance for three extra wins, two against ranked foes, which is something other conferences have been doing for a while, I believe. That'll mitigate at least some of the rankings hit Stanford usually takes.)
Of course, 7 was a bit too low for Stanford this year, and I think 4 would have been a more appropriate ranking/seeding. Of course, that would have put them in line to play the 5 seed in the quarters and the 1 seed in the semis as opposed to being on the 2 vs. 3 side of the bracket. I thought Florida and Vanderbilt were the scariest teams to face coming into this tournament, so I'm happy they're on the other side of the draw.
[/quote]
thanks for the thoughtful comments. I guess what I was thinking is, if the committee should be deciding which are the best teams going into the tournament. If they are basically going to rely on the computer rankings for that, then they should improve the computer rankings. I would support them using the computer rankings as info on the overall season for each team if they would then make reasonable adjustments. Vanderbilt was the hottest/best team going into the tournament so moving them from their #2 computer ranking to a #4 seed makes no sense to me.
I don't think the ITA should build anything into the ranking system to account for Stanford's postseason dominance. Rather, the committee should look at its history of underseeding Stanford, often grossly as in last year, and either adjust its own performance (by understanding how the ranking algorithm is biased) or adjust the computer ranking algorithm so that teams like Stanford are not under-ranked for playing a weaker schedule. Pepperdine is another team this year that was probably underseeded because of its softer schedule.
You're right about basketball in that injuries are not factored into the NCAA's RPI rankings (and probably not in most or all computer rankings either), but the right to be in the tournament field and the seeding within the field are both determined by the selection committee (except for automatic qualifiers by conference championships)



![[Image: Internet_dog.jpg]](https://upload.wikimedia.org/wikipedia/en/f/f8/Internet_dog.jpg)