This is the first report for the 2026 season showing simulated end-of-season ratings for teams, conferences, and regional playing pools. The report is based on the actual results of games played through Sunday, August 23, and simulated results of all scheduled games to be played after August 23 as well as simulated conference tournament games.
As a starting point for determining simulated results, I used the ranking method described in the preceding article (2026 Article 7) to rank all teams. I then assigned ratings to teams based on the historic average rating for each rank level.
Next, I assigned result probabilities to each game -- a win, a loss, and a tie probability. To do this, I (1) identified each team's rating, (2) determined the difference between the two teams' ratings, (3) adjusted the difference to take home field advantage into consideration, and (4) used a history-based result probability chart to assign a win, loss, and tie probability to each team. I used these probabilities as the game result for purposes of the simulation. Thus for a game, Team A would show X.XX% win, Y.YY% loss, Z.ZZ% tie as its game result and opposing Team B would show Y.YY% win, X.XX% loss, Z.ZZ% tie.
I then ran NCAA RPI calculations for all teams based on the actual results of games played and simulated results of all games not yet played.
Once I had done this, I had simulated full-season NCAA RPI ratings (as well as Balanced RPI ratings) for all teams. I then used those NCAA RPI ratings to re-determine the results of games not yet played (thus replacing the initially used ranks and ratings described in Article 7). This gave me current full season simulated ratings and rankings following week 2, and related data.
As a test, for the 549 games already played I compared how well the predicted results using these ratings and rankings compare to the actual results of the games. I used the results of home teams (or for the few neutral site games, the teams whose names are first in the alphabet) as the basis for evaluation. The home teams actually won 51.2% of their games, lost 29.4%, and tied 19.4%. This compared to predicted results of 50.2% wins, 29.3% losses, and 20.6% ties. Thus although some individual game results may have varied significantly from predicted results, overall predicted results came very close to matching actual results. This suggests that over a significant number of games, this is a reasonable method for producing simulated full season rankings.
The results of this process are in the Excel workbook 2026 RPI Report as of 8.23.2026. To download the workbook, use the following steps:
1. Click on the workbook link.
2. The link takes you to a Google spreadsheet, which you will not want to use. On the Google spreadsheet, to the left, click on File, which will bring up a drop down menu.
3. In the menu, click on Download, which will bring up another drop down menu.
4. In the drop down menu, click on Microsoft Excel (.xlsx). This will download the Excel workbook.
The following explanations are for the material in the workbook.
TEAMS
The first page of the workbook is for Teams. I will use the top 10 teams in the simulated full season rankings to illustrate the workbook's material on teams.
This shows the first group of columns on the Teams page. This year, I have revised the way I have divided teams among regions. A main reason for this was to recognize that the Autonomy (Power 4) conferences to a great extent no longer are regional and that their teams tend to compete against each other. As a result, I have created the Autonomy region, consisting of the ACC, Big 10, Big Twelve, and SEC. I also have created the Middle, Mideast, North, South, Southeast, and West regions, with each conference placed in one of those regions. I show the conference assignments below in the section on Regions.
The simulation includes simulated conference tournaments; and the NCAA Tournament Automatic Qualifier column shows the teams that the simulation projects as the conference tournament winners.
For the NCAA Tournament, teams with more losses than wins cannot receive at large positions. The NCAA Tournament Disqualified Due to Losses Greater Than Wins shows, with a "1," the teams disqualified from an at large position based on the simulation.
The NCAA RPI Rank column shows teams' simulated ranks. Next to it, the NCAA Strength of Schedule Contributor Rank column is important to understanding the RPI. Using Alabama as an example, its NCAA RPI rank is #10. Thus you would think, if you were to play Alabama, that within the strength of schedule portion of your own NCAA RPI's calculation, you would get credit for playing the #10 team. You won't. As the NCAA Strength of Schedule Contributor Rank column shows, you only get credit for playing the #39 team. This is a result of the way the NCAA RPI formula computes Strength of Schedule and is a formula defect about which I have written extensively.
I developed the Balanced RPI specifically to fix that NCAA RPI formula defect. The Balanced RPI Rank and Balanced RPI Strength of Schedule Contributor Rank columns show that the Balanced RPI does not have this problem. They also show how the fix changes the ranks.
Continuing to the right on the Teams table, the next columns look at each team's opponents' average strength:
The first two columns bunch conference and non-conference opponents together. Using Stanford as an example, its Opponents Average NCAA RPI Rank is 58. As above, one would think its Opponents Average NCAA RPI Strength of Schedule Contributor Rank likewise would be 58, but it isn't, instead it is 88. In other words, the NCAA RPI formula, because of how it computes teams' strengths of schedule, significantly understates Stanford's opponents' strength when computing Staford's strength of schedule. If you use the workbook itself, you can scroll down through all the teams and see which teams this hurts and which it helps.
The next two columns look only at a team's conference opponents and the two after that look at a team's non-conference opponents.
The next columns look at teams' good results against Top 50 opponents:
Apart from the NCAA RPI itself, results against Top 50 opponents one of the significant factors the Women's Soccer Committee considers in its NCAA Tournament decision-making process. I have developed a system for scoring good Top 50 results that is highly skewed towards good results against very highly ranked opponents. The above two columns show the teams' scores under my system and their ranks based on those scores.
This part of the Teams table is similar to the above table showing the strength of the teams' opponents, but is based on the Balanced RPI rather than the NCAA RPI. As you can see, teams' opponents' Balanced RPI Ranks and their Balanced RPI Strength of Schedule Contributor Ranks are essentially the same. When comparing teams, this gives a much better picture of each team's strength of schedule Overall, in-conference, and non-conference.
CONFERENCES
The second page of the workbook is for conferences.
The first columns, on the left of the Conferences page, shows the conferences and the number of teams in each conference. Next, it shows the average NCAA RPI of each conference's teams and how the conferences rank based on their average NCAA RPIs. And next, it shows how each conference's teams rank based on their average Balanced RPIs. As you can see, although the NCAA RPI and Balanced RPI have the same ranks for the Autonomy conferences, they have some pretty big differences for other conferences.
The next two columns (columns 2 and 3 in the above table) show the average NCAA RPI rank of each conference's teams followed by the average NCAA RPI rank as strength of schedule contributors of each conference's teams. As you can see, some conference's teams are significantly underrated as Strength of Schedule contributors, some are rated about right, and some are significantly overrated.
The next two columns (columns 4 and 5 in the above table) show the average NCAA RPI rank of each conference's teams' opponents followed by the opponents' average NCAA RPI rank as Strength of Schedule contributors. Again, there are significant differences between the two.
The next four columns (columns 6 through 9 in the above table) show similar conference opponents information broken down between in-conference opponents and non-conference opponents.
The last column shows conferences' ranks using the NCAA Non-Conference RPI.
Continuing more to the right:
These columns show information similar to that in the preceding table but for the Balanced RPI. As you can see, unlike the NCAA RPI, conferences' teams' Balanced RPI average ranks and their average ranks as strength of schedule contributors are essentially the same. They also are essentially the same for conferences' opponents.
These columns show the difference between each conference's opponents' average ranks and their ranks as strength of schedule contributors, first for the NCAA RPI and then for the Balanced RPI. A negative number is the extent to which a conference's opponents are underrated as strength of schedule contributors and a positive number is the extent they are overrated.
REGIONS
I have assigned conferences to regions within which their teams tend to play their games, as follows:
Autonomy: ACC, BigTen, BigTwelve, SEC
Middle: Horizon, MidAmerican, Missouri Valley, Ohio Valley, Summit
Mideast: AtlanticTen, BigEast, Colonial, Ivy
North: AmericaEast, MetroAtlantic, Northeast, Patriot
South: American, Southland, Southwestern, SunBelt
Southeast: AtlanticSun, BigSouth, ConferenceUSA, Independent, Southern, United
West: BigSky, BigWest, MountainWest, PacTwelve, WestCoast
I first will show columns from the Regions page that match those on the Conferences page, without re-explaining what they show. Then I will show some additional information from the Regions page with explanations.
Of particular note, the West region's teams play a very high proportion of their games in-region and except for games against Autonomy opponents play almost in isolation from the rest of the country. The North region's teams are in a similar situation with the Mideast, although to a lesser extent.
This shows the proportion of each region's in-region games projected to be ties, which one can consider as a measure of in-region parity. For the last few years, the NCAA's RPI formula has treated tie games as 1/3 of a win when calculating a team's Winning Percentage, rather than the 1/2 of a win it used previously. The table suggests that this change will hurt the Autonomy teams the most, followed by the Mideast teams. This is something the Women's Soccer Committee is monitoring to see whether it wants to continue with the 1/3 of a win part of the RPI formula or revert to 1/2 of a win.
Finally, on the far right of the table, this is similar to the right of the Conferences table. Again, a negative number means that the rating system, within its formula, underrates the region's teams' opponents as Strength of Schedule contributors and a positive number means it overrates them.
No comments:
Post a Comment