In this article, I discuss conference scheduling and how it relates to the NCAA RPI. In particular, the article focuses on strong conferences and their numbers of in-conference as compared to non-conference games. The article uses the Big 10's 2026 scheduling as a case study.
This article is based on teams' actual results through October 11 and projected results of games not yet played, including conference tournament games. Because of that, the data here about the current season are "provisional." At the end of the season, I will update and publish the final data for this case study.
NCAA RPI Background
Because of the way the NCAA constructs its RPI, one should expect that a strong conference with a high proportion of in-conference games will be at a rating and ranking disadvantage, as compared to other strong conferences with lower proportions of in-conference games. The reasons for this are:
1. The Importance of a Team's Opponents' Winning Percentages. The NCAA RPI basic formula is:
(WP + 2*OWP + OOWP)/4
WP is the team's Winning Percentage, computed as:
(Wins + 1/3*Losses)/(Wins + Losses + Ties)
OWP is the average of the team's opponents' WPs (against teams other than the team itself)
OOWP is the average of the team's opponents' opponents' WPs
In the basic formula, the 2*OWP + OOWP portion is referred to as the Strength of Schedule component. Thus overall the formula is a combination of two basic components: the team's Winning Percentage and its Strength of Schedule.
Although it is not immediately apparent from the RPI formula, each of the two basic components has an effective weight of 50%. You can find an explanation of why this is so at the RPI for Division I Women's Soccer website on its RPI: Formula page, under the heading Calculation of RPI. The NCAA publicly has agreed that these are the effective weights of the two components.
The Strength of Schedule's 50% effective weight is split 40% Opponents' Winning Percentage and 10% Opponents' Opponents' Winning Percentage. Thus a team's Opponents' Winning Percentage (40%) is almost as important as the teams own Winning Percentage (50%) and its Opponents' Opponents' Winning Percentage (10%) is a relatively minor contributor. The linked page also cxplains why these are the effective weights of the Strength of Schedule elements. (The NCAA, so far as I know, has not publicly acknowledged and may not what people to know these are the effective weights, but I am confident the NCAA knows they are.)
The key piece of information here, for purposes of this article, is this: A team's Opponents' Winning Percentage comprises 40% of the effective weight of what makes up its NCAA RPI rating.
2. The Effect of In-Conference Games on a Conference Team's Opponents' Winning Percentage. As a hypothetical, suppose I am in a conference that plays a full round robin. In relation to my in-conference opponents' winning percentages, suppose I play conference opponents A and B. If A has beaten B, then when I play A, its Winning Percentage will include its win over B. Likewise, however, when I play B, its Winning Percentage will include its loss to A. Or, if the two have tied, each's winning percentage will include the tie. This means that the effect, from all of my in-conference games, is that my Opponents' Winning Percentages will net out at 0.5000 -- for every conference opponent win there is a matching conference opponent loss and for every conference opponent tie there is a matching conference opponent tie.
Since all conferences do not play full round robins and since there are conference tournaments in which all teams may not participate, the net effect from all in-conference games does not net out at exactly 0.5000. In general, however, the effect of all of my conference games will net out to an Opponents' Winning Percentage of right around 0.5000. (This likewise will be the case for the net effect from in-conference competition on Opponents' Opponents' Winning Percentage.)
A result of this is that if all conferences played only conference games, the NCAA RPI would not be able to distinguish the conferences from each other.
3. The Importance of Non-Conference Games in Distinguishing the Conferences from Each Other. The way in which the NCAA RPI distinguishes the conferences from each other is through non-conference games. If I am a conference team with an excellent winning percentage in non-conference games, that excellent winning percentage plus my in-conference winning percentage is what I contribute to each of my conference opponent's Opponents' Winning Percentage. Thus it is to all of my conference opponents' NCAA RPI advantage for me to have an excellent winning percentage. The cumulative effect of all of my conference's teams' non-conference winning percentages is by far the main factor that distinguishes my conference from other conferences, from an NCAA RPI perspective.
4. The Importance of the Proportion of Games That Are Conference and Non-Conference. Suppose that my conference plays half of its games in-conference and half non-conference. Then the half that are in-conference will set my Opponents' Winning Percentage at about 0.5000. The half that are non-conference will cause an adjustment to that 0.5000 based on how good or poor my other conference members' winning percentages are in their non-conference games. If you assume I am in a strong conference whose members have high non-conference winning percentages, my games against them will cause my Opponents' Winning Percentage to be well above the 0.5000 that comes from my conference games.
As an alternative, suppose again I am in a strong conference whose members have high non-conference winning percentages, but that we play two-thirds of our games in-conference and only one-third non-conference. The kind of effect of the high non-conference winning percentages will be the same as in the prior example, but because of the lower proportion of non-conference games, the distance my games against my conference opponents will pull my Opponents' Winning Percentage above 0.5000 will be smaller.
Thus if you have two equal-in-strength conferences whose teams have equal non-conference winning percentages, but one conference plays half its games as non-conference whereas the other plays a third as non-conference, the conference that plays half its games as non-conference will have overall higher Opponents' Winning Percentages than the conference that plays a third of its games as non-conference and thus will have higher NCAA RPI ratings. Although the ratings will make it look like the conference with half its games as non-conference is the stronger conference, that in fact is not the case. Rather, its appearance as stronger is simply a reflection of how the RPI formula computes strength of schedule combined with the difference between the conferences' proportions of in-conference and non-conference games.
What Do This Year's Data Show?
1. What Are the Autonomy (Power) Conferences' Proportions of Conference and Non-Conference Games? Here are the numbers of in-conference regular season games and the numbers of conference tournament games for the Autonomy (Power 4) conferences:
Big 10: 12 regular season games; 8 tournament participants, 7 games; total of 85 games. 12.78 games average per team.
Big 12: 11 games; 8 participants,7 games; total of 73 games. 11.88 games average per team.
SEC: 10 games; 12 participants, 11 games; total of 66 games. 11.38 games average per team.
ACC: 10 games; 6 participants, 5 games; total of 60 games. 10.59 games average per team.
Here are the numbers of non-conference games:
Big 10: 5.94 games average per team.
Big 12: 7.44 games average per team.
SEC: 7.13 games average per team.
ACC: 7.29 games average per team.
These numbers translate into the following proportions:
Big 10: 68.2% in-conference, 31.8% non-conference.
Big 12: 61.5% in-conference, 38.5% non-conference.
SEC: 61.5% in-conference, 38.5% non-conference.
ACC: 59.2% in-conference, 40.8% non-conference.
Based on these numbers, the Big 10, with its highest proportion of in-conference games, would need the highest winning percentage in its non-conference games, of the four conferences, in order to end up with an overall Opponents Winning Percentage to match theirs. And, the ACC with its lowest proportion of in-conference games, could get away with the poorest non-conference winning percentage.
2. What Are the Conferences' Actual Winning Percentages in Non-Conference Games? Here are the Autonomy conferences' actual winning percentages in non-conference games:
Big 10: 64.5%
Big 12: 65.8%
SEC: 69.6%
ACC: 76.9%
With these non-conference winning percentages and the conferences' proportions of non-conference games, it should be no surprise that the projected end-of-season conference NCAA RPI ranks are ACC #1, SEC #2, Big 12 #3, and Big 10 #4.
3. What Are the Conferences' Opponents' Strengths? The most powerful consideration for the Women's Soccer Committee when making its NCAA Tournament seeding and at large selection decisions is teams' NCAA RPI ranks. Another powerful consideration, however, is teams' results against NCAA RPI Top 50 opponents. And, within these results are results against non-conference Top 50 opponents. Here are this year's projected team average numbers of non-conference Top 50 opponents for the four Autonomy conferences:
Big 10: 1.72
Big 12: 1.31
SEC: 1.88
ACC: 1.47
This results in the following proportions of non-conference games against Top 50 opponents:
Big 10: 29.0%
Big 12: 17.6%
SEC: 26.3%
ACC: 20.2%
In addition to this, here are the average NCAA RPI ranks of the conference teams' non-conference opponents:
Big 10: 125
Big 12: 135
SEC: 129
ACC: 142
Thus according to the NCAA RPI, the Big 10 is playing the most difficult overall non-conference schedule and in addition, of its non-conference games, plays the highest proportion of Top 50 opponents. The Big 10's playing the highest proportion of Top 50 opponents, of its non-conference games, is not surprising since it plays roughly the same number of Top 50 opponents as the other Autonomy conferences but plays fewer total non-conference games.
What this means, in relation to the above NCAA RPI Background discussion, is that given its 12 game in-conference schedule, the Big 10's non-conference scheduling has been the opposite of what it needed to be in order to offset how the Strength of Schedule component of the NCAA RPI formula works. Its non-conference schedule needed to be weaker than the schedules of the other Autonomy conferencces and in fact, it has been stronger.
4. How Much Does This Affect NCAA RPI Ratings? Here are how each Autonomy conference's teams' actual performances so far this season relates to their expected performances based on their current NCAA RPI ratings. The numbers are the difference between the conference teams' actual winning percentages and what the NCAA RPI ratings say their winning percentages should have been if the ratings are correct. A positive percentage means the conference teams' actual winning percentages have been better than their expected winning percentages:
Big 10: 6.5%
Big 12: 3.8%
SEC: 5.8%
ACC: 4.6%
As expected, all of the conferences' performances to date are better than their ratings say they should have been. In other words, the NCAA RPI underrates all of the Autonomy conferences.
Of these, the Big 10 is the most underrated. In light of the above data and discussion, this is what one should expect. Because the Big 10 plays the highest proportion of in-conference games, one would expect the NCAA RPI, because of how it calculates Strength of Schedule, to underrate it more than it underrates the other Autonomy conferences.
Putting it All Together
The Big 10, by playing a 12-game in-conference regular season, has reduced the number of non-conference games its teams are able to play. Given the structure of the NCAA RPI, with its high valuation of opponents' winning percentages, this puts the Big 10 in the position of needing its teams to have higher non-conference winning percentages than the other Autonomy conferences teams, in order to equally compete with them in the rankings. From this perspective, if the Big 10 wants to play a 12-game in-conference regular season, then from an NCAA RPI perspective it needs to schedule weaker non-conference opponents than the other Autonomy conferences.
On the other hand, if the Big 10 wants its teams to have good Top 50 non-conference results equivalent to those of the other Autonomy conferences, it needs to be scheduling the same number of Top 50 non-conference opponents as the other Autonomy conferences.
In other words, the Big 10's decision to play a 12-game in-conference regular season, coupled with the NCAA RPI's high emphasis on opponents' winning percentages, has put the Big 10 teams in a "damned if you do and damned if you don't" position. They have to chose between being hurt by playing weaker non-conference schedules than the other Autonomy conferences or being hurt by playing stronger non-conference opponents on a per-non-conference-opponent basis.
In future years, the simple solution for the Big 10 is to stop playing a 12-game in-conference schedule. In fact ideally, in relation to the Autonomy conferences, they all would play the same number of in-conference games and would have equally sized conference tournaments. This would put them all on an equal footing from an NCAA RPI formula perspective.
NCAA Tournament Seeding and At Large Selection Effects
In terms of NCAA Tournament seeds and at large selections, does it matter?
Remember, the problem I have described comes from the way the NCAA RPI calculates strength of schedule, specifically its high emphasis on Opponents' Winning Percentage. The Balanced RPI does not have this problem and is able to eliminate most, if not all, of the effects of conferences having different proportions of in-conference and non-conference games. Thus if you compare the Women's Soccer Committee's likely NCAA Tournament seeding and at large selection decisions based on the NCAA RPI to what they would be if the Committee were using the Balanced RPI, you can get a picture of how much the Big 10's having a 12-game conference schedule is costing it from an NCAA Tournament perspective.
The second half of my 2026 Article 18 shows the difference between the Big 10's NCAA Tournament prospects under the NCAA RPI and if the Committee were using the Balanced RPI, based on all teams' actual results through Week 9 of the season and their projected results through the balance of the season. Although the difference shown in Article 18 is provisional and we will not know the actual difference until the end of the season, Article 18 suggests that the Big 10's playing a 12-game in-conference season, given that the Committee uses the NCAA RPI to rank teams, is going to significantly diminish how the Big 10 will fare in the seeding and at large selection process.
This is the first year in which I have done these kinds of calculations. Given that, I am apprehensive about whether the extent to which the Big 10's NCAA Tournament seeding and at large selection prospects will be diminished due to its 12-game in-conference schedule ultimately will end up as great as the numbers currently suggest. This will be something to track as we go through the coming weeks.
No comments:
Post a Comment