Southern Cal looked good against Florida yesterday, completely bossed a talented gator squad in the second half and scored three goals to win 3-0. That win followed their tie with a very good Florida State team three days earlier. Trojans are impressive. Wake Forest might be one of the most underrated teams in the nation. They battled Tennessee to a 1-1 tie a week or so ago (good game) then yesterday made a first-half goal (a blast from 25 yards by Baley Feist) stand up to take a 1-0 win against Penn State in Winston Salem. I won't say the win was entirely deserved as Penn State had 12 shots on goal for the game against 4 for WF, including three or four great chances in the first half, but poor finishing by the Nittany Lions and a couple of excellent saves by the WF keeper were the difference. The WF keeper almost bungled the game in the final seconds: Penn State sent the ball into the box and the WF keeper tried to punch it away; however, she mis-hit the ball, which bounced directly to a Penn State player in the box who looked to have an open net, but a WF defender made the play of the game, getting a foot in front of the opposing player at the last moment to flick the ball away. Outstanding play.
As per the previous poster, Tuesdays is the day for the RPI. However, if the NCAA follows past practice, it won't publish RPI rankings until the Tuesday after the fifth week of the season. That would be September 18.
Forgive my late reply, but I disagree on two counts. Firstly, everyone who makes the NCAA tournament as a non host should have a drivable road game. I know that goal is not 100% fulfillable, but the players and fans do deserve it, and I think it is why they went to 32 first round sites instead of 16, and I think it's why sometimes you see a team leap several RPI spots from say, 49 over 39 to make the cut. The marginal differences between two teams in the RPI range of that would likely not overcome the geographic factor... and your prospect of having some of those regional first round match-ups be 17 states away just for the sake of balance seems ludicrous to me. Secondly, let's face it, this is still a non revenue sport, and always will be. I don't want to lose the "amateur"-ish aspects of the college game, and I'm not sure why you'd suggest otherwise. I don't think people are going to pay even $30 to see a first round game just because it's an 8/9 draw to see who will get the privilege of going to the #1 overall. (Whereas in basketball/football, these scenarios could actually be much more lucrative to the point of almost feasible.) The example of why your proposal is errant, is that they're never going to send the 64th team all the way across the country to lose to the number one overall, and once again no one would pay $30 required to fund such an event (airfare, extra lodging costs). If it works out that way sometimes fine, but regional pods make total sense financially and practically. It's good enough to me that the #1 team overall will face someone between 12th and 16th best in their side of the bracket in the first round, with home field advantage through the first three rounds. This current system is fine and fair, for all participants and you don't have to worry about some seeded team getting a bad draw because all things considered, the tourney is five rounds, and whatever apparent disadvantages some seeded team gets based on their geographical match-ups may be leveled out over the course of the first two weekends, most times.
I don't think teams leaping several RPI spots has anything to do with geography. There are other very good explanations. And, differences of several spots are not very meaningful. Neither the RPI nor any other rating system can distinguish among teams at that level of accuracy. Beyond that, I agree with your post. When women's soccer can sell NCAA Tournament rights for a hundred million dollars, that's when we may see the seeding of 64 teams. With the 32 first round sites, however, the Committee is talking about the possibility of seeding 32 teams. They already effectively do it (although seeding teams on "pods" instead of 1 through 32) in the first round site selection process.
The simulation system I use intentionally is based strictly on the RPI. In order to project "beginning of season" ratings for teams, after evaluating lots of options, it appears that the best RPI-based formula for beginning of season ratings is (Next Year's Computer-Projected Rating Based on a Trend Line for the Last Eight Years + Average of the Last Two Years' Ratings)/2. It's a crude measure, but its ratings work better for the coming year than any other RPI-based system I can come up with. So, I determine beginning of season ratings using that formula for every team. Then, after I've got the full schedule for the season, I apply those ratings, adjusted for home field advantage, to every game to determine the game outcome. With all of the simulated game results, my computer then determines teams' ratings and ranks. One problem with the system is that it has trouble with teams changing directions. Princeton was an example, this year it had a significant change from its trend. So, as is the case for any detailed pre-season prediction, you have to take it with a big grain of salt -- which is what I always say about the pre-season simulation. Interestingly, however, if you look at how teams ended up in their conference standings at the end of the season, the simulation's pre-season standings were only minimally less accurate than those of the conference coaches and of Chris Henderson's highly informed pre-season standings.
Could maybe find predictive drop-offs with loss of seniors/coaching change? Would be some real data mining but if you could find out a team was losing over half their starters that would mean something, I would think.
That's something I'm sure the coaches do, if only intuitively, and it's something Chris Henderson does overtly. That my system uses only historic ARPI ratings and has an overall prediction success close to theirs is what I find quite interesting. It reinforces my suspicion that the accomplishments of programs over the long term are more important than most people think. Ideally, a system like I use, overlaid with the kinds of data you suggest, might be the best.
Princeton is still very good. They only lost 3 games, and of those, only the loss to Brown could be considered a surprise/bad loss. I would expect that's the one that knocked down their RPI a fair bit, but they still finished #24.
Yeah I would imagine using something like Goal Impact's set up of rating every player would be the best way to eliminate noise but... that's a lot of work for a lot of players who won't go professional : /