What the 148 means is that over the years since 2007, there has been at least one team, ranked #148 as of the completion of week 6 of the season, that ended up getting an at large selection. So I treat #148 as the outside edge of teams for which it seems an at large selection is a possibility. It is unlikely, of course, but not impossible. (As I recall, it was an SEC team that went on a tear in conference play.) I like #148 because it gives an idea of how fluid RPI ratings and ranks still are even though we are entering week 7. [NOTE: There are 31 conferences with automatic qualifiers, so there are 33 at large positions.]
Do you ever wonder exactly what data the members of the Women’s Soccer Committee are looking at during their NCAA Tournament seeding and at large selections process? Of course, the NCAA publishes RPI ranks weeky at the NCAA.com Division I Women’s Soccer page. In the menu to the right of Women’s Soccer, click on Rankings. This will bring up a page showing the United Soccer Coaches most recent rankings. Click on the United Soccer Coaches box and then, in the drop down menu, click on NCAA Women’s Soccer RPI. This will bring up a page showing the Adjusted RPI most recent rankings. That is nice. But, there is far more available to the Committee. And thanks to great work by the NCAA staff, we all now are able see what the Committee can see. To get your own look at the current information: Use this link to go to the RPI Summary Reports for 2022-23 Division I Women’s Soccer: The NCAA publishes detailed data reports weekly starting on completion of the fifth week of the season, typically on Monday morning and including within its data base all games completed since the preceding Sunday. At the top center of the page the above link takes you to, you can specify the date as of which you want to see the data reports. The linked page shows the teams in RPI rank order. It includes data for each team broken down in a variety of ways. Of particular interest on this page are columns that show the Adjusted RPI rating (as distinguished from rank) of each team and the Conference ARPI average rating and rank of each team’s conference. And, additional columns show the team’s record against teams within different rank ranges. And, as an added bonus, you can click on the title of each column to sort the page based on the data in that column. But wait ... that is not all ... In the team column, each name is a link. Click on a team’s name and you will go to a statistics page for that team. The statistics page includes the team’s schedule, including the results of the games played so far. In the schedule, click on the game score and you will go to the box score for that game. Plus, from this page you can navigate to pages with lots of stats about the team, the players, and the coach. But wait ... there is more ... Back on the page with the teams in RPI rank order, each team rank number is a link. Click on a team’s rank and you will go to what is called a Team Sheet for that team. This is a page full of information about the team that relates to factors the Committee must consider in making its decisions. It also has links to the statistics page for each opponent that team played and to the box scores for each game the team played. The Team Sheets, to my mind, probably are the most important data pages in relation to Committee decisions about seeds and at large selections. But wait ... there is even more ... Back again on the page with the teams in RPI rank order, in the row at the top that begins with Summary Report, there is a series of links to other pages: Ranking Report, Non-Conference Ranking Report, Conference Ranking Report, and Conference Non-Conference Ranking Report. If you are truly interested in the Committee decision-making process, becoming familiar with these resources is a must!
At the RPI and Bracketology for D1 Women’s Soccer blog, I have three new posts with: (1) Current actual RPI ratings and ranks and additional data for each team, including indications of teams within the current ranges for potential seeds and at large selections, (2) Simulated end-of-season Adjusted RPI ranks, (3) Evaluations of teams for each of #1, 2, 3, and 4 seeds and at large selections based on Committee historic decision patterns, plus simulated conference champion automatic qualifiers using the actual results of games played through October 2 and simulated results for games not yet played. Plus, in the NCAA Tournament piece, I have a small discussion about Alabama and its #1 seed prospects. And, in the Simulated end-of-season RPI ranks piece, I have a small discussion about why Florida State remains in the 20s in the simulated end of season ranks and what it will take for it to move higher up in the ranks.
At the RPI and Bracketology for D1 Women’s Soccer blog, I have three new posts with: (1) Current actual RPI ratings and ranks and additional data for each team, including indications of teams within the current ranges for potential seeds and at large selections, (2) Simulated end-of-season Adjusted RPI ranks, (3) Evaluations of teams for each of #1, 2, 3, and 4 seeds and at large selections based on Committee historic decision patterns, plus simulated conference champion automatic qualifiers using the actual results of games played through October 9 and simulated results for games not yet played. In the NCAA Tournament evaluation piece, I have included some discussion about issues the Committee would have to consider if the simulated results were the actual results at the end of the season. Hopefully, this will help in gaining an understanding of how the Committee decision-making process might work.
Looking over your post and am trying to figure out what the difference is between RPI, Adjusted RPI and Adjusted NCRPI?
The basic RPI, sometimes called the normal or unadjusted RPI, is determined by a formula that includes winning percentage, opponents’ winning percentages, and opponents’ opponents’ winning percentages. Once the software computes the unadjusted RPIs of all teams, there are bonuses and penalties added or subtracted from the ratings, to determine Adjusted RPIs. The bonuses are for wins or ties against teams ranked 1-40 in the unadjusted RPI and lesser bonuses for wins or ties against teams ranked 41-80. (The bonus amounts also have gradations based on whether the game was home, away, or at a neutral site.) The penalties are a mirror image with the tiers being the bottom 40 teams and the next to bottom 40 teams. Just as there is an unadjusted RPI covering all games, there is an unadjusted Non-Conference RPI determined by the same basic formula but based on non-conference games. The purported purpose of this is to properly rate conferences (not teams) in relation to each other. There likewise are bonuses and penalties in the same amounts, with the same tiers, but the tiers use the URPI ranks and not the UNCRPI ranks. If you really want to understand these in detail (and are a glutton for punishment), there are full explanations at the RPI for Division I Women’s Soccer website.
From the just-published 2022 Pre-Championship Manual: SEEDING AND BRACKETING The Competition Oversight Committee has approved a bracketing pilot for the Division I Women’s Soccer Championship. Fifty percent of the bracket will be seeded in four groupings of 1-8. Published seeding will be in four sets of 1-8 and the pairings will honor a 1-8 seed assignment. The committee will identify the top 32 teams. The top 16 teams will be identified in rank order and assigned a seed grouping from 1 to 4. The balance of the ranked teams, in turn, are assigned to one of the remaining seed groupings (i.e., 5 to 8). Once the seed assignment is finalized, it remains unchanged while placing the teams into the championship bracket. First and second-round conference matchups will be avoided. This is a big change and, in my opinion, a very good one.
At the RPI and Bracketology for D1 Women’s Soccer blog, I have three new posts with: (1) Current actual RPI ratings and ranks and additional data for each team, including indications of teams within the current ranges for potential seeds and at large selections, (2) Simulated end-of-season Adjusted RPI ranks, and (3) Evaluations of teams for each of #1, 2, 3, and 4 seed pods (seeds #1 through #16) and at large selections based on Committee historic decision patterns, plus simulated conference champion automatic qualifiers, plus -- new this week -- teams for each of the #5, 6, 7, and 8 seed pods (seeds #17 through #32), using the actual results of games played through October 16 and simulated results for games not yet played. In the NCAA Tournament evaluation piece, I have included some discussion that may help give insight into some issues the Committee may have to consider in making its decisions.
Current RPI has it UCLA, UNC, Alabama, ND then FSU, then ARkansas, UVA and Duke My question is why is Alabama so high. Their loss is to Miami at 62 in the current RPI. Their tie was to Utah at 57. Their 4 best wins are #6 Arkansas, #13 BYU, #15 USC and #18 Clemson. Meanwhile FSU's losses are to #s 2 UNC and #4 ND. Our ties are to #15 USCe and #56 Auburn. Our 4 best wins are #7 UVA, #8 Duke, #12 Pitt and #18 Clemson. My non wonky analysis shows that our best wins are comparable. Our losses and Ties w/the exception of Auburn are much better. I thought losing to #62 would carry more negative weight than apparently does. We've played 4 of the top 10 and 7 of the top 20. Alabama has played 1 of the top 10 and 4 of the top 20. ?????
The RPI formula considers three elements: winning percentage [Effective weight 50%] average of opponents’ winning percentages (slight simplification) [Effective weight 40%] average of opponents’ opponents’ winning percentages [Effective weight 10%] (The effective weights are as of the end of the season, so they are a little different now.) As you can see, the RPI does not care who the wins, ties, and losses were against. There are minor bonuses added to the number produced by the RPI formula, based on the ranks of opponents against whom you have good wins or ties, but they bunch opponents ranked 1-40 together and 41-80 together and ordinarily have minimal effects on team rankings. When the Committee gets to working on the NCAA Tournament bracket, it takes a detailed look at the ranks of opponents and the results against them, which is why the seeding does not necessarily match the rankings.
At the RPI and Bracketology for D1 Women’s Soccer blog, I have three new posts with: (1) Current actual RPI ratings and ranks and additional data for each team, including indications of teams within the current ranges for potential seeds and at large selections, (2) Simulated end-of-season Adjusted RPI ranks, and (3) Evaluations of teams for each of #1, 2, 3, and 4 seed pods (seeds #1 through #16) and at large selections based on Committee historic decision patterns, plus simulated conference champion automatic qualifiers, using the actual results of games played through October 23 and simulated results for games not yet played. I have the feeling things this year will look a little different than usual to the Committee. This could be due to the aftermath of the 2020 Covid disruption, increased parity, and possibly the elimination of overtimes. It is going to be interesting to see what the Committee does.
At the RPI and Bracketology for D1 Women’s Soccer blog, I have three new posts with: (1) Current actual RPI ratings and ranks and additional data for each team, including indications of teams within the current ranges for potential seeds and at large selections. This includes some new data allowing you easily to compare team ARPI ranks with their ranks by the RPI formula as strength of schedule contributors to their opponents, which ranks can be quite different although they should not be. It also allows you to compare team ARPI ranks with ranks by a revised RPI formula that would remove the ARPI rank to strength of schedule contributor rank discrepancies as well as remove the ARPI discrimination against teams from strong conferences and geographic regions, (2) Simulated end-of-season Adjusted RPI ranks, and (3) Evaluations of teams for the #1, 2, 3, and 4 seed pods (seeds #1 through #16), for the new #5, 6, 7, and 8 seed pods (seeds #17 through #32), and for at large selections, all based on Committee historic decision patterns, plus simulated conference champion automatic qualifiers, using the actual results of games played through October 30 and simulated results for games not yet played. Although last week I thought quite a few team profiles were going to look odder than usual to the Committee, the current data shows an overall picture that looks pretty similar to past years. The four new seed pods will be of particular interest to me this year, as the Committee has not done them previously.
At the RPI and Bracketology for D1 Women’s Soccer blog, I have posted my final assessment of what the Committee might do for seeds and at large selections, if it follows its historic patterns.
Nice work. Looks like the only discrepancy between your selections and Chris Henderson’s are that you have Texas A&M in over Wake Forest. Not sure how Texas A&M gets in with no Top 50 wins while WFU has 2. Also head to head with WFU, WFU wins out. WFU beat Clemson and Georgia and tied Auburn while Texas A&M lost to Georgia, tied Clemson and beat Auburn. I think ACC gets 10 teams in, especially given how strong the conference was this year.
With both being in the category of teams that have no profile aspects that say they always would have gotten in and likewise no profile aspects that say never, it would be consistent with past Committee history for either of them to get in or be out. There are 10 individual factors that are equally adept at picking the right selections from among the last remaining candidates and I simply picked one of them that is relatively simple. But none of the 10 factors is always right although the total system is right 95 percent of the time. This could be one of the misses.
Here is a quick summary of how the system I use did, compared to the actual bracket. Eventually, I will do a full report. The system: #1 seeds: correctly picked all four. #1 through #4 seed pods (16 teams): without regard to which pod, correctly picked 15 of the teams. In terms of distribution among the pods, missed a number of them, all by one pod. This is not a surprise. The system had Georgetown as a #4 seed, but it did not get seeded. It had Northwestern as not seeded, but it got a #4 seed. #5 through #8 seed pods (16 teams): without regard to which pod, correctly picked 12, which is about as expected. In terms of distribution among the pods, it was a bit hit and miss, which is not surprising. The system had UCF and Brown as #7 seeds and Xavier and Arizona State as #8 seeds, but they did not get seeded. The four teams getting seeded in their places are West Virginia and Mississippi State with #7 seeds and NC State and Portland with #8 seeds. At large: correctly picked all but one. Wake Forest got an at large position and Arizona did not. The system had Arizona not part of the final bubble group (barely). If it had been in the final bubble group, it would have been out and Wake Forest in. What this means overall is that the Committee held pretty consistently to its historic patterns. It also suggests that the change to the no overtime rule did not have a significant effect on how the Committee makes its decisions.
Slightly off topic and I'm a little out of the loop due to the NWSL madness but why did NCAA do away with overtime this year? Is it a permanent rule now and did the coaches approve of it to protect athletes from burnout? Did this rule apply to men's college soccer as well?
My understanding is it is something that the coach organization wanted and is related to player safety concerns and the unequal competition that occurs if the first match on a weekend is overtime for one team in the second match of a weekend but not for the other team in the second match. It is permanent, so far as any rules are permanent. And it applies to all NCAA soccer competition regardless of level and to both women and men.
Since I am interested in how different approaches do in making pre-season predictions, I track the pre-season predictions of where teams will finish in their conference standings. The three I track are those by the conference coaches, Chris Henderson’s, and those based on my pre-season assigned ratings derived from team historic trends. The three systems take different approaches. For the coaches, they rely on their detailed knowledge and experience of each others’ teams. For the Chris Henderson predictions, he relies on a series of data-driven metrics combined in a formula. For mine, I rely on team historic ranking trends converted to ratings. Here is how the three did this year: Average difference between team actual end-of-season conference standing and pre-season predicted standing: Coaches: 2.44 Chris Henderson: 2.44 CPT: 2.61 Team ranks exactly right: Coaches: 41 Chris Henderson: 45 CPT: 37 #1 ranked teams that become NCAA Tournament Automatic Qualifiers: Coaches: 9 Chris Henderson: 10 CPT: 9 #1 ranked teams that become conference regular season champions: Coaches: 9 Chris Henderson: 8 CPT: 8 It looks like tweaks Chris Henderson did to his system have worked well, as on the average difference metric, the coaches have been the best over the past few years, but their sets of predictions now match up. Mine always have lagged slightly. One of the things that stands out to me is that my purely trend-driven system performance is reasonably close to that of the detailed analyses that the coaches perform (consciously and unconsciously) and that CH produces consciously. This has been the case over the past few years and continues to surprise me.
It would be interesting to see the teams with the largest RPI deltas from pre-season to end-of-season.
As I read one bracket, you've got 2nd-round matches next Friday followed by a 3rd-round match on Sunday--one day of rest! Can that be right? Why wouldn't you play the 2nd-round matches on Thursday so the teams would have 2 days of rest? I thought player welfare was important--that's why the NCAA said it eliminated overtime. Even worse, a lower-seeded team that wins its 2nd-round match on Friday would have to travel that night or on Saturday (its supposed rest day) to the home field of the higher-seed and play its 3rd-round game the next day--Sunday. Completely unfair.
You are right that the second and third round games appear to be Friday-Sunday. I am pretty sure the coaches lobbied hard for them to be Thursday-Sunday, but I guess they lost that battle, although they did win it for the semi-finals and finals. However, the second and third round matches are played in groups of four at the site of the highest seed in the group. So there only is one trip, not two.