Below in this week's bonus article, I show a predicted NCAA Tournament bracket based on the Women's Soccer Committee's historic patterns (1) with the Committee using the NCAA RPI as its rating system and (2) as if the Committee were using the Balanced RPI rather than the NCAA RPI. I also include Chris Henderson's projected bracket so that you can compare all three. The Henderson bracket projection will not quite give an apples-to-apples comparison, as his data includes a couple of games from Monday, October 27.
The NCAA RPI and Balanced RPI predictions are based on the actual results of games played through Sunday, October 26, and predicted results of games not yet played including conference tournament games. The predicted results are based on teams' current NCAA RPI ratings, in other words what the results should be if all teams perform exactly in accord with their ratings as adjusted for home field advantage.
In addition, I am including a case study showing why the Committee likely would include Saint Louis as an at large team if using the NCAA RPI but instead would include Kansas State if using the Balanced RPI and how the difference relates to the way the NCAA RPI computes strength of schedule.
PROJECTED BRACKET COMPARISON
The following table shows projected brackets. Scroll to the right for a key:
Saint Louis or Kansas State Case Study
In this week's table, there is only one team that gets an at large position using the NCAA RPI -- Saint Louis -- that does not get one using the Balanced RPI, and one team that gets an at large position using the Balanced RPI -- Kansas State -- that does not get one using the NCAA RPI. So the question for today is: Why the Saint Louis/Kansas State switch? The key is in how the NCAA RPI computes the teams' strengths of schedule.
NCAA RPI Formula, Rank
Saint Louis
RPI Rank: 27
Opponents' Average Rank/Rank as Strength of Schedule Contributors: 127/139
Conference Opponents' Av Rank/Rank as SoS Contributors: 115/114
NonConference Opponents' Av Rank/Rank as SoS Contributors: 147/180
Kansas State:
RPI Rank: 54
Opponents' Average Rank/Rank as Strength of Schedule Contributors: 102/122
Conference Opponents' Av Rank/Rank as SoS Contributors: 56/91
NonConference Opponents' Av Rank/Rank as SoS Contributors: 174/171
In the above information, the important thing to note is that Saint Louis' opponents' ranks as strength of schedule contributors to its RPI are 12 positions poorer than their actual RPI ranks whereas Kansas State's are 20 positions poorer. In other words, although the RPI appears to be understating Saint Louis' strength of schedule, it is understanding Kansas State's strength of schedule by significantly more.
Here are the comparable numbers for the Balanced RPI:
Balanced RPI Formula, Rank
Saint Louis
RPI Rank: 46
Opponents' Average Rank/Rank as Strength of Schedule Contributors: 140/140
Conference Opponents' Av Rank/Rank as SoS Contributors: 139/139
NonConference Opponents' Av Rank/Rank as SoS Contributors: 141/142
Kansas State:
RPI Rank: 48
Opponents' Average Rank/Rank as Strength of Schedule Contributors: 85/85
Conference Opponents' Av Rank/Rank as SoS Contributors: 41/41
NonConference Opponents' Av Rank/Rank as SoS Contributors: 155/155
Note here that the teams' opponents' RPI ranks and ranks as strength of schedule contributors are essentially the same, rather than having the differences of the NCAA RPI.
The result of using the Balanced RPI is that Kansas State moves up in the rankings from 54 to 48. And the bigger result is that Saint Louis moves down from 27 to 46. Why does Saint Louis drop so far? Because there are more teams than Kansas State that are in the same opponents' strength of schedule situation as Kansas State and when those teams' opponents' strengths of schedule are corrected, they move ahead of Saint Louis in the rankings.
In addition to the change in RPI rank, there is another change using the Balanced RPI that is a prime contributor to Kansas State displacing Saint Louis as an at large team. That change is in the teams' good results against Top 50 opponents, which when paired with the RPI is the prime driver of the Women's Soccer Committee's at large decisions (as demonstrated by the Committee's historic decision patterns in relation to the data it sees).
NCAA RPI Formula, Top 50 Results (Good Results)
Saint Louis
#20 Dayton Win Away, Points 144
Total Top 50 Results Points 144, Total Points Rank 60
Kansas State
#13 Colorado Tie Away, Points 576
#35 UCF Tie Away, Points 12
Total Top 50 Results Points 588, Total Points Rank 44
Note: Under the NCAA RPI, Saint Louis' RPI is better than Kansas State's, but Kansas State's Top 50 Results Score is better than Saint Louis'. However, Kansas State's Top 50 Results are not enough to overcome Saint Louis' better RPI.
Balanced RPI Formula, Top 50 Results (Good Results)
Saint Louis
#43 Dayton Win Away, Points 4
Total Top 50 Results Points 4, Total Points Rank 96
Kansas State
#8 Colorado Tie Away, Points 2880
#23 UCF Tie Away, Points 90
Total Top 50 Results Points 2970, Total Points Rank 33
Note: Under the Balanced RPI:
(1) Dayton's rank is poorer, for reasons similar to why Saint Louis' rank is poorer. As a result, Saint Louis' Top 50 Results score and rank are poorer.
(2) Colorado's and UCF's ranks are better, for reasons similar to why Kansas State's rank is better. As a result, Kansas State's Top 50 Results score and rank are better.
(3) The net effect is that Kansas State, rather than Saint Louis, gets an at large position.
Thus once the RPI's defective way of computing strength of schedule is corrected, Kansas State gets an at large position and Saint Louis does not.
I am confused. If Saint Louis wins at Dayton, Saint Louis would be the A10 AQ, not an at large selection. That is because that would be the A10 Tourney Final. Dayton would need the at large selection. Both teams deserve to be in the NCAA Tourney. I Would like to see more deserving non football four programs make it. The deck is already stacked against them in many ways with resources. They can’t even get home and home games with many football four programs
ReplyDeleteIf the A10 teams perform in accord with their current ratings, Dayton will win the Tournament and be the AQ. That is the scenario my case study addresses, which of course could be different than how things end up.
ReplyDeleteA good number of people would like to see more non-Power 4 teams in the NCAA Tournament -- although there automatically will be 26 of them as Automatic Qualifiers. It seems to me that what those people are saying is that they think team strength as demonstrated over the course of the season not only should not apply in relation to Automatic Qualifiers but also should not apply to a bunch of other at large positions. I think that is a reasonable position to take, although I do not agree with it, even though I am not a Power 4 fan. The NCAA current rules, however, require at large positions to be based on team strength as demonstrated over the course of the season.
I was reading your top 50 result for Saint Louis as having a win away at RPI #20 Dayton. It’s listed as a good result. They haven’t done that yet. If they do that it will be in the conference tourney final to be the AQ as opposed to being an at large. Maybe I am reading that wrong.
ReplyDeleteBased on current ratings and home field advantage which would go to Dayton, I have the conference tournament final as a tie, not a win by either team. In other words, although Dayton currently has the better rating, it is close to Saint Louis' rating and home field advantage to Dayton would not be enough to make a Dayton win the best prediction. In that case, my system awards the pk result to the better rated team, which in that case would be Dayton. So Dayton gets the Automatic Qualifier position and Saint Louis is competing for an at large position.
ReplyDeleteI can see why this might have been confusing.
I see. Thanks. Enjoy your work.
ReplyDelete