2025 D1 Women's Soccer Tournament Discussion [Pre-, PBP, Post-] [R]

Yes, that could factor into it. Also consider that because of the inherent bias in RPI that @cpthomas identified, the SEC ends up being “over seeded”. If my visual inspection is correct, only three seeded teams - who are also home teams - lost in the first round. All are from the SEC

TN (3) - UNC
MS St. (6) - Lipscomb
UGA (8) - Ohio St

There were some close calls
- Utah State had four chances to beat BYU (5) in PK’s. Once if their goalie made a save, and three times if their shooter scored. The PK highlights are worth a look.
- Northwestern lost in PKs at Alabama (8) whose backup goalie saved 3 shots.
- Louisville (6) beat KY in PKs

 

I wonder if the remoteness of Tallahassee compared to DC had anything to do with it. The #2 seed hosts two games if they win the first round. If their #3 seed wins in the first round they travel their next two possible games if their #2 seed wins.

 

Physical play and kick ball (sorry Direct Play0 eventually catchup with you. SEC does a lot of it and come tournament time they get exposed.

 

I’d hope that was not a factor for FSU. Non-conference SOS is much more relevant for teams from weaker conferences. Not for teams that play each of the top 3 teams in the country in conference play. Our (FSU’s) overall SOS in Massey is 6th.

 

Here are the result probabilities for the second round games, based on teams' NCAA RPI ratings as adjusted for home field advantage. The first % column is wins, the second losses, the third ties. The totals at the bottom suggest the better seeded teams, in the first column, will win 9 games, lose 3, and be tied in 4 after regulation time with those 4 games decided by golden goal or PKs.

 

Apparently FSU has to plead to get non-conference games. Not sure that I would accept the challenge if I were an upper 3rd level team knowing that chance of picking up a win is minuscule. Thanks to Kansas for taking the challenge!

 

If you're suggesting the Committee might have taken travel implications into consideration in giving Florida State a #3 seed, I think that would not be the case. The Committee seeds teams without regard for travel implications.

 

If I'm reading Chris's table right, Georgetown had a Strength of Schedule of 6 and FSU had an SoS of 20.

EDIT: Posted this without seeing the later on discussion of SoS, whoops

 

Just call up the Big East. Georgetown flew down to Nashville to play Vandy (in addition to bussing down to Charlottesville). Xavier flew to Charlottesville. UConn flew down to College Station. Creighton flew to Durham. They would all provide a stiffer test than Stetson or FIU.

 

So FSU played the #1 team (loss) 2 team win and loss, 3 team loss, 7 team tie #15 team win, # 25 team win, 28 team win.
Georgetown played #2 UVA tie (their best result IMO), #18 twice a win and a loss
I suppose us playing Stetson and FIU negates all that.

 

If we look at RPI (which has its many flaws, and obviously it isn't controlling on the committee since West Virginia has an RPI of 13 and got a 7 seed, but it is illustrative), the reason why Georgetown (12) was within striking distance of Florida State is because, in addition to UVA (2) and Xavier (18), the Hoyas played Vanderbilt (4), South Carolina (37), St. Mary's (42), UConn (44), James Madison (51), and Old Dominion (52).

I'm sure Chris will correct me if there's no way this would be a factor, but it may just have come down to the fact that Georgetown has 5 more wins than FSU, mostly a function of playing 4 more matches.

 

Alright. I'm a committee of one to determine which team I believe was more worthy of a #2 seed - FSU or GTown. Below, I attempted to get both FSU and GTown results versus playoff teams, with seeding included, since the committee created the seeds. I have also included "bad" results against non-playoff teams. For playoff teams, their seed is included. Otherwise, I include RPI - another committee metric. Please let me know if I mistated a result or missed a game I should have included.

*** FSU (RPI #10)
* Good Results
W @ Vir (#1)
W Louis (#6)
W WF (#7)
W UNC (RPI #25)
T @KS (#3)
??? T Vir (#1) (lost in PKs) ??? Not sure it's good, but don't believe it's bad to tie a #1 at home

* Bad Results
L @ ND (#1)
L Stanford (#1)
T @NC State (non-playoff, RPI #115)
T Cal (non-playoff, RPI #46)


*** GTown (RPI #12)
* Good Results
W @Xavier (#6)
T @Vir (#1)
W SC (RPI #37)

* Bad Results
L @ Vandy (#1)
L Xavier (#6) Neutral
T ODU (non-playoff, RPI #52)
L SM (non-playoff, RPI #42)


Analysis
-----------
Results that "cancel":
1) L @ND and L @Vandy - both AWAY and both #1 seeds
2) T Cal and T ODU - both HOME RPU #46 and #52 respectively
3) FSU: (T @KS and W UNC) GTown (T @ UVA and W SC)

What's left?

*** FSU (RPI #10)
* Good Results
W @ Vir (#1)
W Louis (#6)
W WF (#7)
??? T Vir (#1) (lost in PKs) ??? Not sure it's good, but don't believe it's bad to tie a #1 at home

* Bad Results
L Stanford (#1)
T @NC State (non-playoff, RPI #115)

*** GTown (RPI #12)
* Good Results
W @Xavier (#6)

* Bad Results
L Xavier (#6) Neutral
L SM (non-playoff, RPI #42)

Bad results are similar, but FSU has noticably better remaining good results w/o including the tie with UVA.
I'll also note FSU (RPI = 10) has a higher RPI than GTown (RPI = 12).

FSU earns my #2. That and $5 will get you a cup of coffee.

 

I have no horse in the Florida State as compared to Georgetown seeding race, but here are some interesting numbers for comparison:

Florida State

RPI Rank #10
Opponents' average RPI Rank 75
Opponents' average Strength of Schedule contribution rank 108
Record 10W 2L 4T
ACC rank #1
Conference regular season standing 5
Conference tournament standing 5.5 (lost in quarterfinals)
*Actual winning percentage less expected winning percentage based on RPI rating 4.6% (i.e., Florida State did better than its RPI rating said it should have done)
​

Georgetown

RPI Rank #12
Opponents average RPI Rank 109
Opponents' average Strength of Schedule contribution rank 132
Record 15W 3L 2T
Big East rank #5
Conference regular season standing 1
Conference tournament standing 2
*Actual winning percentage less expected winning percentage based on RPI rating -1.1% (i.e., Georgetown did slightly more poorly than its RPI rating said it should have done)

*Counting ties as 1/3 of a win, per the current RPI formula
​

Comments

The RPI is not precise enough for there to be a big difference between a #10 and #12 rank. This is why there have been #2 seeds ranked as low as #13.

The average opponents' rank for a #2 seed is ~73. The poorest is ~129. Both Florida State and Georgetown had strengths of schedule within the historic range for #2 seeds.

In 2022, the NCAA changed the formula for the Winning Percentage element of the RPI, which has a 50% effective weight in the overall RPI calculation. Prior to the change, in the Winning Percentage formula a tie counted as half of a win. The change reduced the tie value to a third of a win. Florida State had 4 ties and Georgetown had 2. So the change hurt Florida State this year, relative to Georgetown. [Personal comment: In my opinion, this was an uninformed and irresponsible change by the Committee. The RPI was a poor rating system before the change and the change made it worse.]

As I said in an earlier post, I had Florida State as a #2 seed. Georgetown, however, also was within the range for a #2 seed. If I am going to make a criticism, it is not of the Committee, which could have gone either way. (Other than TCU, all of the #2 seeds and Florida State, had profiles that said "yes" to a #2 seed but also said "no," which means they all had profiles the Committee never had seen before.) My criticism is of the RPI and in particular of: (1) the differences between the teams' opponents' average RPI ranks and the opponents' average ranks as RPI formula strength of schedule contributors and (2) the devaluation of ties from half a win to a third of a win. These two things devalued Florida State in relation to Georgetown this year. I also would criticize the coaching community for not raising holy hell about the RPI formula and for going along with changing the value of ties.

As a matter of interest, here are the teams' ranks by the four rating systems:

NCAA RPI Florida State 10 - Georgetown 12
KPI 8 - 12 (which the Committee can use as a backup to the NCAA RPI)
Massey 6 - 9
Balanced RPI 9 - 22​

 

Out of curiosity, I Iooked to see which aspects of Florida State's profile appear to have done it in, for a #2 seed. Its RPI rating was in the range of teams that never have gotten #2 seeds, barely. Its NonConference RPI rating, on the other hand, was more inside the no #2 range. It looks like these, combined with where it finished in the conference regular season and tournament standings, were "red lights" for a #2 seed.

This is not as a basis for comparison to Georgetown. Rather, it suggests that Florida State's profile this year is not what one would expect for a #2 seed, based on past history.

In the past, I would be saying that Florida State needs to have higher ranked teams on its non-conference schedule, if it wants a very high NCAA Tournament seed. In the new NCAA Division I sports landscape, however, that is harder to do than it was in the past, for travel cost regions. To be able to play higher ranked teams, you have to be willing to enter into home-away agreements, which means you traveling to the higher ranked school as well as their traveling to you. That means significant travel expenses, and less money is available for that than was avaliable in the past, either available to you or available to your potential highly ranked opponent.

Assuming Florida State wants to be competing for high NCAA Tournament seeds, its Athletic Department will need to pony up enough money to cover the travel expenses needed to schedule a significant number of highly ranked opponents. We'll get to see if the Department cares enough to do that.

 

But, why? They played more games against playoff teams and more games against higher ranked teams than Georgetown. Why should it matter whether they are in conference or not?

 

I mean, we were a number 1 seed seven straight years from 2018-24. The last three of those under Pensky and a similar scheduling philosophy to 2025. I hardly think this will be some kind of impetus to overhaul FSU's or Pensky's thinking with respect to non-conference scheduling.

And for the record, as a fan, I'd love for us to play more of those games, top teams from other conferences, home, away, whatever.

Would help the non-conf if Florida ever got their act together.

 

I put this in the FSU thread but given all the talk about FSU vs. Georgetown for the #2 seed I'll put it here as well.

Lipscomb upset MSU.

There were a lot of people upset over FSU getting snubbed for a 2 seed (no shade, I was one of those people).

However, now that we know the draw you can make a pretty compelling argument that the Noles actually lucked out with the 3 seed.

Had the committee made FSU the 2 seed and Georgetown the 3 seed the Noles would have been able to stay at home but would have had to play West Va (#16 Massey) next for the right to face Georgetown (#10 Massey) in the Sweet 16. Now they get to play Lipscomb (#130 Massey) and Georgetown and WVU have to fight it out in the other game.

I know that staying at home is a major benefit but I think that I would sacrifice home cooking for the chance to get to the Sweet 16 without playing a team in the top 100 (Samford is ranked 101 Massey).

 

To be clear about my post about FSU's non-conference schedule, what I am talking about is the Committee's historic seeding patterns. The coaches whose teams are competing for NCAA Tournament seeds are well aware that the Committee decision follow patterns. What those patterns are is one of the areas I work on and coaches have access to what the patterns are. I'm simply saying that FSU's non-conference schedule this year left them outside of the #2 seed range based on some of the Committee's patterns. People can complain and criticize, but it doesn't change the fact that the Committee has patterns and coaches need to schedule accordingly.

 

It's poor form in data analysis to take prior models of how teams are ranked, and blindly use that model as the basis for future rankings. Seedings act as predictions, and also create a bias in favor of the seeds due to getting home games. What should be done is all seeding criteria used should be correlated against the results to determine which criteria are most predictive and which should be discarded or modified. I’m thinking that “not making the conference tournament”, or “placement in the conference based on regular season record” will be significantly different in terms of achievement in the ACC (for women’s soccer) versus other conferences. Last year was a significant example as ND (#3) and Stanford (#4) went on the road and beat #1 and #2 respectively. ND finished 6th in the ACC and lost in the quarterfinals. Stanford finished 8th - dead center of the conference and missed the ACC playoffs. This is just an example. But what I’m really criticizing is the process - as I understand it. It’s not like the committee posts their reasoning for their decisions - unless I missed it.

 

The Committee does not explain its decisions -- and I don't blame them for that. Imagine how much more controversy it would cause -- and, it probably wouldn't be possible since different Committee members might have different reasons for their decisions.

The system they use is interesting. For at large selections, by NCAA rule they must be based on evaluating teams' performance over the course of the season as demonstrated by game result data that the NCAA provides to the Committee, in relation to specific NCAA-mandated factors the Committee must consider. The selections are not based on how "good" teams are, except there is a little "out" for the Committee that lets it look at performance over the last eight games. For seeds, although the Committee uses the same data it uses for at large selections, it is not required to use those data or to apply the at large factors. Nevertheless, the Committee's decisions over time have been consistent with using the data -- while at the same time allowing for ranges within which the Committee can go one way or the other

I am entertaining the idea that a good predictive model for how teams will do in the NCAA Tournament is the median of teams' ranks (using a variant of the NCAA RPI) over the last seven years, translated into ratings and taking home field advantage into consideration. This may seem peculiar, but it suggests that by the end of each season, as coaches have been able to work with their players, teams tend towards their historic mean -- or median, in this case. That model worked pretty well last year and so far is working pretty well this year at predicting tournament results. If it continues to work well, then your approach would suggest it would be a good basis for seeding.

 

There's no need to accept the conference versus nonconference argument. A game is a game, and, playing in the ACC, FSU had a tougher overall schedule than GTown. Period.

 

I've been thinking about this and whether it reflects actual practice anywhere in professional sport seeding.

In tennis, with maybe a single exception, tournament seeding is based on past performance and not attempts to be predictive. (The exception is Wimbledon, which is played on grass and the seeding gives that fact particular consideration although it still may "retrodictive" rather than "predictive" seeding.) In US professional baseball, football, and basketball, end-of-season tournament seeding is based on past performance. In European soccer, who gets advance to the "higher" league competition from the different federations is based on past performance. In World Cups, how teams are distributed among the groups is based on past performance (as measured by FIFA rankings).

In college conference tournaments, seeding is based on past performance. In college national championships, which teams are at large selections is based on past performance. How at large teams are selected is based on past performance, by rule. How teams are seeded may be an exception, at least the committees have some leeway, but seeding still largely is based on past performance. And those who post here certainly consider past performance, just look at the arguments they are making about why Florida State should be preferred over Georgetown, which are based on past performance.

In other words, if teams in a sport competition are seeded anywhere on a predictive basis, it is a rarity. It seems likely the reason for this is that using a retrodictive approach to seeding seems more objective than a predictive approach. It also says you have to earn a seed by demonstrating through your performance that you deserve it. At least that is what the NCAA and lots of other sports' governors seem to think. So you are advocating for something that is very different than the norm.

 

But, don’t each of the sports you mention above come up with their rankings in a transparent, rigidly defined, repeatable way? In other words, anyone take those specific sport’s formula or rules and come up with the same seeding/ranking. That’s fundamentally different than NCAA soccer seeding.

 

So, which matchups is everyone excited for?

I mainly follow Big 10, Big East, certain Big 12 teams and the mid-major conferences, so for me I will be watching:

CU vs Xavier
Georgetown vs WVU

And then my night cap is going to be TCU-Memphis, mainly because I haven't gotten to watch much of Memphis.

Then obviously, I think both Friday matchups are going to be good. Open to changing my viewing plans based on other thoughts.

 

I definitely think all eyes will be on the TCU regional this weekend. So many storylines. Is Memphis legit? Will UNC continue their warpath as an unseeded defending champion? Is the Big 12 for real this year? Plus any implications for the coaching carousel.