RPI for 2008

I had thought that Massey was discontinuing doing college soccer ratings this year, but he isn't. He now has ratings out covering games through October 5. To see his ratings (as well as conference ratings), use the following link:

http://www.mratings.com/rate.php?lg=csocw

Massey's system takes into account game location as well as goal differential. It starts the season with some historic data figured into the ratings, but by the end of the season has extruded all historic data and relies only on this year's data.

UPSoccerFanatic wrote:I had thought that Massey was discontinuing doing college soccer ratings this year, but he isn't. He now has ratings out covering games through October 5.

I think this happened last year. I think Massey doesn't post results early in the season because he feels they have no meaning.

This may be information overload, but I've been working on my weekly report comparing the strength of the different regions, based on inter-regional games. I have been concerned about whether the teams not included in the average RPIs by region were skewing the results that I have been providing, and in particular about whether they have been skewing the results in favor of the West Region. So, this week I've done an analysis of the teams not included in the data. Here's the info, including the current average RPIs by region, for you to make of what you will.

First, there are 13 teams that have no inter-regional games. Here are the number of teams in this group by region, plus the average RPI (all games) of these teams by region:

Central 0.3532 (7 teams)
Great Lakes 0.3659 (1)
Mid Atlantic none
Northeast none
Southeast 0.4293 (2)
West 0.4864 (3)

Second, the way my program works, there are a good number of teams for which there are not enough inter-regional games in their playing "pod" to compute an RPI based only on inter-regional games. Here are the number of teams in this group by region, plus the average winning percentage in inter-regional games (RPI Element 1) of the teams for each region. I've used average RPI Element 1 because that's the one number I can compute for each of these teams:

Central 0.3860 (22 teams)
Great Lakes 0.5463 (19)
Mid Atlantic 0.4419 (10)
Northeast 0.5000 (11)
Southeast 0.4410 (14)
West 0.6028 (15)

With all of the above teams excluded from the average RPI by region calculations, here are the average RPIs by region for the remaining 214 teams:

Central 0.4916
Great Lakes 0.4826
Mid Atlantic 0.4828
Northeast 0.5108
Southeast 0.4852
West 0.5467

Finally, here are the percentages of teams excluded as described above from the last table, by region:

Central 50.9%
Great Lakes 33.3%
Mid Atlantic 20.8%
Northeast 26.8%
Southeast 26.2%
West 35.3%
Total, All Regions 32.7%

As you can see, and has been the case previously this season as well as for the season last year, the average RPI by region numbers indicate that the West is the strongest region by significant margin. What my additional vetting of those numbers indicates, however, is that the numbers not only don't overstate the West's strength, they understate it. This is because the teams from the West not included in those numbers are significantly stronger than the teams not included from the other regions.

Note: I just checked the Massey ranking and it is now calculated through Oct. 12th.

UPSF, (and anyone else who wants to respond) - the way Portland quickly dropped from #1 to #3 in the RPI rankings gave me pause. It has dawned on me just how dicey this RPI can be.

It's not anything like a poll, which takes votes from a certain group of "experts" and then ranks the teams according to how many points they are given by the group. Polls have flaws, of course. Like "East Coast Bias" among other possibilities.

The RPI is even scarier though. In a poll, a media writer from Rhode Island can see Boise State has an 11-1 record. (American Football). However, the writer can look at the team's strength of schedule and get a fairly good idea what Boise State has accomplished.

The RPI, however, is much more complex. The Pilots beat Gonzaga 4-0 and drop from #1 to #3. On the surface, it makes no sense at all. But with UPSF's explanations of how the system works, it make total sense.

But is the RPI system really fair? The Pilots and a few other teams are in a position where they have earned the right to control their destiny at least to a point. But like so much "fairy dust", poof!!! We have no control whatsoever over most of the RPI system, except our own won-loss-tie record, and what teams we scheduled that year are doing. And we really don't know how the teams on our schedule will do in a given year. We could conceivably, I think, win out and not get a #1 seed? Correct?

And that would be absurd...

You bring up an interesting point, FANatic. With the RPI system they use for women's soccer, it totally relies only on wins and losses. Strength of schedule stuff is of course worked in as well, but at the end of the day the core stat everything is built off of is wins and losses.

Now (I can't believe I'm going to say this, but here it is), I actually like that the BCS uses the human polls as well, because it can add at least a little bit of subjectivity to the rankings. I mean, if the BCS operated like the women's soccer RPI, it's likely that teams like Utah, BYU and Boise State would be a lot higher when the rankings come out this weekend. Now, I realize that the BCS system is f'd up, but that's mainly because it doesn't have a playoff. If it operated in such a way that the top 8 teams made it to a playoff, I actually think that it would provide a pretty accurate list of the top eight teams in the country.

FANatic wrote:It has dawned on me just how dicey this RPI can be. ... But is the RPI system really fair? The Pilots and a few other teams are in a position where they have earned the right to control their destiny at least to a point. But like so much "fairy dust", poof!!! We have no control whatsoever over most of the RPI system, except our own won-loss-tie record, and what teams we scheduled that year are doing. And we really don't know how the teams on our schedule will do in a given year.

I agree with your comments, but I've also tried to put myself in the NCAA's position. They are completely averse to using polls because of the opportunities for bias and lobbying and appearances of unfairness. They refuse to use weighted statistical systems based on historic performance (even if better predictors of game outcomes) for philosophical reasons. If one accepts those limitations for purposes of analyzing their position, they're pretty stuck. To illustrate, supposed they have two bubble teams at the end of the season that have identical win-loss-tie records. The NCAA must pick one of those two teams for an at large position. What are they to do? All they can do is look at their strength of schedule and say, "Well, yes you both have the same record, but Team A had that record against a tougher schedule. So if Team A had played the weaker schedule that Team B played, then it's likely Team A would have had a better record. Thus Team A gets the selection." That makes a certain amount of sense. The problem is with how the RPI identifies strength of schedule. It's a "blind" calculation system. By "blind" I mean that all it looks at is the average winning percentage of a team's opponents and a team's opponents' opponents, without regard for who any of those opponents or opponents' opponents are. In theory, Team A's and Team B's playing groups could be completely separate with no cross-group games. So, Team A's playing group could be NCAA Division 1 teams and Team B's playing group could be grade school kids. Assuming both groups' end-of-season results followed a standard bell curve distribution, the two groups would have identical RPIs. Thus the top NCAA Division 1 team would end up with the same RPI as the top grade school team. The assumption of the NCAA is that there are enough inter-group games for it to be able to treat all its teams as part of a single functional group, so that unlike the Division 1/Grade School example, the RPI indeed can fairly rank all teams in a single group. However, that is not the case. There are not enough inter-regional games for the NCAA to be able to do this. I proved that last year in a way that I think cannot be refuted.

If all the regions are of equal strength, the problem I just described would not matter. Each region, over the long term, would have a standard bell curve distribution of results and the top teams from each region would make it into the tournament and into the seeds in proportion to the numbers of teams in the different regions. The problem is if the regions are not of equal strength. In those cases, the RPI tends nevertheless to treat the regions as of equal strength. In other words, it treats Team A's .6900 RPI achieved with Strong Region as a home base the same as Team B's .6900 RPI achieved with Weak Region as a home base. In fact, however, it is harder to achieve a .6900 RPI in Strong Region than it is in Weak Region, so Team A's .6900 RPI actually indicates it is a better team than Team B notwithstanding Team B's .6900 RPI.

So, in real lilfe, are the regions of equal strength or not? This is the significance of my yesterday's post comparing regions' average RPIs based on inter-regional games. Those comparisons indicate that there are regional strength differences. In particular, the West Region is significantly stronger than the others. The same was true last year, as confirmed by both the Massey and SoccerRatings systems. So, the RPI discriminates at this point against teams from the West Region. It treats a .6900 RPI of a West Region team the same as a .6900 RPI of a team from another region, when it shouldn't.

Both Massey and SoccerRatings have a similar problem with there being relatively few cross-regional games. However, they do not use the "blind" system the RPI uses. Both Massey and SoccerRatings use a different type system based on the fundamental concept that if Team A beats Team B and Team B beats Team C, then it is more likely than not that Team A will beat Team C. Thus those two systems "see" who the different opponents are. Because of this ability of these systems to "see," I expect their ratings to be able to take different regional strengths into account, limited only by the limited number of cross-regional games. Thus I would expect their systems to have a greater proportion of teams at the upper area of the ratings than does the RPI. I haven't done exact counts, but I believe they in fact meet this expectation.

Note, however, that to the extent that the Pilots are competing with other West Region teams for seeds, this is not an issue. The issue is in the competition with teams from other regions.

Here are average RPIs by conference, based only on inter-conference games. The order of information is: RPI conference rank, conference, conference average unadjusted RPI, and Massey's conference rank.

1 Pac 10 0.6072 1
2 ACC 0.6067 2
3 Big 12 0.5783 4
4 WCC 0.5671 5
5 Big 10 0.5657 3
6 SEC 0.5615 7
7 Big East 0.5553 6
8 Ivy 0.5482 9
9 Big West 0.5394 8
10 Mtn West 0.5346 10
11 Colonial 0.5325 11
12 CUSA 0.5314 12
13 Atlantic 10 0.5159 13
14 Southern 0.4935 18
15 Horizon 0.4851 14
16 Mid American 0.4803 15
17 Sun Belt 0.4723 24
18 Atlantic Sun 0.4668 25
19 Metro Atlantic 0.4633 22
20 America East 0.4624 21
21 Missouri Valley 0.4608 16
22 Big South 0.4589 26
23 Big Sky 0.4561 19
24 Patriot 0.4545 20
25 WAC 0.4529 17
26 Northeast 0.4510 23
27 Independent 0.4404 30
28 Southland 0.4301 29
29 Summit 0.4240 27
30 Ohio Valley 0.4177 28
31 United 0.3951 31
32 Southwestern 0.3425 32

UPSF, I think I have an idea about the apples to oranges comparison problem you describe when there are differing strengths of regions. Perhaps there can be a points benefit/penalty for being in a stronger or weaker region based upon that regions performance in the championship tournament the previous year. There would still be a year lag in the analysis but if you look at trends they tend to take a few years to develop ie. the Pac 10 has been consistently strong for quite a while now as has been the ACC. This system would at least give the RPI a performance based criteria which addresses the problem of comparing region strength.

Of course any change in the way the RPI works is going to be frought with politcal potholes dug by cronically weaker regions who benefit year in and year out from the system simply "resetting to 0" at the beginning of each year. This is where the Albyn Jones type of historical analysis tends to be more accurate and transperant that the NCAA RPI at least at the beginning of the season.

If the NCAA is in fact looking for more fairness in choosing the seeds and the at large picks this type of approach should be considered. However, I suspect the NCAA is as much a political body as it is a rational body and I doubt the liklihood of any "real change" taking place. So in close I guess my point is the best that can be done is to point out the problems with the system and hope someone takes notice.

Thanks for all the work on explaining the RPI to me and everyone else. Without your work I can honestly say I would have absolutely no clue how the RPI worked and thus would not even have enough information to establish an opinion. Thanks again.

Auto Pilot wrote:
(UPSF)...Thanks for all the work on explaining the RPI to me and everyone else. Without your work I can honestly say I would have absolutely no clue how the RPI worked and thus would not even have enough information to establish an opinion. Thanks again.

Amen Auto. UPSF's "teaching" method of explaining this mountain of RPI data is nothing short of amazing. A fews weeks ago I didn't know enough about RPI to even begin to write the post I wrote above.

It is really just wonderful that UPSF can provide this critical data that at least gives us an idea where the NCAA is coming from. Obviously, there is no perfect system. (Football's BCS standings are controversial and upsetting to people nationwide every year.)

The fact that I could even understand RPI enough to write my post above is a strong testament to UPSF's knowledge, skills, problem-solving ability and a talent in teaching through words that is quite remarkable indeed.

I wouldn't be surprised if UPSF's topics on this subject are being read by a larger and wider audience as time passes. I am serious when I say, perhaps he is establishing a regional and even beginning to establish a national following that is being read by coaches, teams, fans and other reader's alike.

Based on the fact that this post has had almost 1300 views (even if it is 6 pages long) makes me think you are right about the audience FANatic.

And don't forget, I think UPSF posts about the RPI on a nationwide soccer site as well. UPSF's explanations and analyses are really very well written and I wouldn't be surprised if they're referred to by fans all over.

UPSF posts on Big Soccer, the worlds biggest soccer board if you want to find his contributions look for posts by CP Thomas. His crusade to enlighten the world about the RPI is already legendary. Check it out.

Albyn Jones' SoccerRatings now are out for games through October 12. To see them, use the following link: http://soccerratings.com/index.php?title=Women%27s_NCAA_Division_I_Soccer_Ratings

Here's an interesting comparison of the Jones, Massey, and RPI ranking systems. The numbers are the number of teams from the West Region in the top 47 under each ranking system. I've used the top 47 because #47 was the lowest RPI-ranked team selected for an at large position in the NCAA Tournament last year.

RPI: 11
Massey: 13
Jones: 16

I was curious to see how the RPI rates teams from different regions in comparison to AJ's SoccerRatings and Massey's ratings. Here's the number of teams, by region, in the top 50 in each of the rating systems, for games through October 12:

Central RPI: 9 SR: 7 Massey: 7

Great Lakes RPI: 8 SR: 8 Massey: 8

Mid Atlantic RPI: 8 SR: 8 Massey: 10

Northeast RPI: 1 SR: 1 Massey: 2

Southeast RPI: 12 SR: 9 Massey: 9

West: RPI: 12 SR: 17 Massey: 14

The Geez already has mentioned on the LMU game thread that Texas A&M lost tonight to Kansas 1-0 and Washington State beat Cal 2-1. These both are important games for the Pilots. TAMU was one of the teams competing for one of the four #1 seeds, so their loss is great for the Pilots. Also, WSU is starting to look like a possibility for an at large selection for the NCAA tournament. A couple of more wins in the Pac 10 and they should have a real shot. Think: Home games for Rounds 1 and 2!

In the meantime, the teams the Pilots already have played this year have gone 7-4-1 so far this week. The WCC teams the Pilots have yet to play are 1-1-2, and they are going to be 50-50 no matter what. So this week so far has been pretty good for the Pilots' strength of schedule part of the RPI.

On a different note, sad in some respects, Santa Clara and St Mary's tied tonight 1-1. Santa Clara now is 0-1-1 in the WCC, which almost certainly means they won't win the conference title and get a guaranteed NCAA Tournament berth. And, with this tie, they will have a losing record this year which means that by NCAA rule they cannot even be considered for an at large berth in the Tournament, no matter what their ultimate RPI rank. This makes the Pilots' game with them especially important, as SCU will be fighting like a wounded bear, so ... GO PILOTS!

And, as a side note, Portland State seems to be getting better as their season progresses and so far is in the running in the Big Sky Conference.

Very interesting season!

Stonehouse asked a question a while ago about whether it matters, for the Pilots' RPI, what the results are of other WCC games and I said that basically it doesn't matter since a win and a loss, or a tie and a tie, are added to the Pilots' strength of schedule for every conference game. After more thought about it, and as I've spent time with the bonuses and penalties that are added to the basic RPI to yield the adjusted RPI that the NCAA uses, I realized that what I said is not right.

To understand why, you have to understand the bonus system, which is what will be relevant for the Pilots. A team gets bonuses for high quality wins and ties. The amount of the bonus a team can get from a game depends on three things:

(1) What is the unadjusted RPI rank of your opponent? If the opponent's rank is 1-40, then there is one level of bonus; and if the rank is 41-80, then there is a lower level of bonus. Beyond 80, there are no bonuses.

(2) Did you win the game or have a tie? The bonus level for wins is higher than for ties.

(3) Was the game away, neutral, or home. Away wins/ties are worth more than neutral; and neutral are worth more than home.

In terms of WCC games, it is with reference to the "opponent's rank" factor that conference game outcomes can make a difference. For example, when we play San Diego @ San Diego, we want them to be ranked high enough that they will be in the top 40 of the unadjusted RPI at the end of the season. That way, if we beat them, we will receive the highest bonus award possible. Similarly, with Loyola Marymount, we want them to be at least in the top 80 and, ideally, in the top 40, at the end of the season. So, we want the WCC teams with the best possibility to be in the top 40, and then the teams with the best possibility to be in the top 80, to win their games.

Fun, isn't it, thinking about all these details!

The Pilots' non-conference opponents had a great day today. When added to their results from earlier in the weekend, their total weekend results are 14-5-1. Much better than the 50-50 results from last weekend.

In addition, the Pilots will be adding Pepperdine's and Loyola Marymount's non-conference records into their numbers. (I think this is the better set of numbers to look at, rather than including their conference results.) The two teams' combined records are 10-10-5. Combining these two sets of numbers, I think the Pilots' Element 2 (opponents' winning records) will decline slightly but not much. In addition, the Pilots will be adding the Pepperdine and Loyola Marymount opposing teams' records to the Pilots' opponents' opponents records. Loyola Marymounts' opponents, in particular, have pretty good records; and Pepperdine's are decent. I think the Pilots' Element 3 (opponents' opponents' winning records) will increase slightly. And, the Pilots' Element 1 (Pilots' winning record) will increase from .9167 to .9286.

I agree, by the way with a comment on the Pepperdine game thread that we shouldn't spend too much time worrying about the RPI. It will do what it will do. The key for the Pilots is to keep winning games, one by one. I'll keep providing information, but we need to keep it in perspective.

I've just posted an RPI report covering games through October 19 on the BigSoccer website under the RPI thread. Look for the cpthomas post at about 9:52 am. Here are some of the basics:

1 Stanford .7222
2 North Carolina .7221
3 Portland .7134
4 UCLA .7018
5 Notre Dame .7010
6 Florida State .7003
7 Duke .6705
8 Florida .6695
9 Boston College .6690
10 Oklahoma State .6661
11 Texas A&M .6578
12 Virginia .6554
13 Penn State .6487
14 USC .6482
15 Colorado .6448
16 BYU .6365
17 Texas .6362
18 Washington .6264
19 Rutgers .6235
20 Princeton .6209

27 Long Beach State .6067

29 San Diego .6041
30 California .6038

36 Arizona State .5991

38 Washington State .5982

40 Arizona .5961

48 Loyola Marymount .5891

56 Oregon .5818

In the above list, I've included all the West Region teams from Oregon upward. Last year, the lowest RPI-ranked team to get an at large selection was #47.

Notre Dame dropped from #2 last week to #5 this week, despite winning two games. This reveals an odd quirk of the RPI. So long as Notre Dame remains undefeated, its winning percentage always will be 100%, so Element 1 of its RPI will stay at 1.0000. For all the other teams, whose RPIs are below 1.0000, every time they win a game their Element 1 goes up. Notre Dame dropped in the ranks because its strength of schedule declined and had no increase in its Element 1 to offset that decline.

A lot of matches in the next 2 weeks that have heavy RPI Implications.

Friday 10/24
(01) Notre Dame (21) Rutgers

(07) USC (04) UCLA

(12) Virginia (02) North Carolina

Thursday 10/30

(02) North Carolina (08) Florida State

Friday 10/31

(04) UCLA (05) Stanford

Sunday 11/02

(07) USC (05) Stanford

(04) UCLA (18) California


So by around the 3rd of Nov, we should have a pretty good idea of what the pecking order may look like.

Also bear in mind that UNC has the ACC Tourney in which they will end up facing at the very least one of Virginia, Duke, FSU,

Also Notre Dame has their conference tourney which, as we know anything can happen come tourney time.

In addition to the RPI report in my previous post, here are the average RPIs, by Region, based only on inter-regional (in other words, out-of-region) games:

Central: 0.4746
Great Lakes: 0.4769
Mid Atlantic: 0.4773
Northeast: 0.4810
Southeast: 0.4836
West: 0.5351

(For those who have been following the regional numbers I've been reporting on a weekly basis, I've figured out how to use Excel so that all but the 12 teams with "no inter-regional games" are included in these numbers.)

There are 12 teams that have played no inter-regional games, so they made no contribution to the above average RPIs by Region. To be sure that the exclusion of the 12 teams was not unduly skewing the above numbers, I calculated the average RPIs of those excluded teams, by region, as well as the percentage of each Region's total number of teams that the excluded teams represent. Here are those results:

Central: 7 teams with no inter-regional games. Average RPI of 0.3358. 12.3% of region's teams.

Great Lakes: 1 team with no inter-regional games. RPI of 0.3380. 1.7% of region's teams.

Mid Atlantic: 0 teams with no inter-regional games.

Northeast: 0 teams with no inter-regional games.

Southeast: 2 teams with no inter-regional games. Average RPI of 0.3704. 3.3% of region's teams.

West: 2 teams with no inter-regional games. Average RPI of 0.4684. 3.9% of region's teams.

I believe these numbers for the teams with no inter-regional games suggest that the Great Lakes, Southeast, and West average RPIs by region would be slightly, but not significantly, lower if these teams had had inter-regional games so they could be included. The Central average RPIs would be lower to a more significant degree, although not greatly lower.

Since teams are playing almost entirely intra-conference games at this point in the season, with only a very few exceptions, I am not planning at this point on doing any more average RPIs by conference, based on inter-conference games, until after the NCAA Tournament. The numbers will stay about the same as for the numbers I've provided previously.

The NCAA has released an updated version of the RPI covering games through October 21. This was a previously unannounced release date, but the Division 1 Women's Soccer Committee apparently asked that the RPI be released three times during the season, so this is the second of three releases. My previous RPI results covering the same period are similar, though not identical. The Pilots are at #3 in both versions. The differences most likely are due to my having to guess the bonus/penalty amounts.

To see the NCAA RPI, use the following link: http://web1.ncaa.org/app_data/weeklyrpi/2008WSOrpi1.html

UPSoccerFanatic wrote:The differences most likely are due to my having to guess the bonus/penalty amounts.

Why must you guess? Do they not release their criteria?

dystopia membrane wrote:

UPSoccerFanatic wrote:The differences most likely are due to my having to guess the bonus/penalty amounts.

Why must you guess? Do they not release their criteria?

Seems like it would make sense for the NCAA to release all its criteria, doesn't it? They release all their criteria and all of the RPI formula (sort of) with one exception: the penalty and bonus award amounts. Those are a closely guarded secret.

What we do know is that there are bonuses awarded for wins and ties against teams in the top 80 of the "unadjusted" RPI (which is the RPI before the awarding of bonuses and penalties). We know that the bonuses are more for wins and ties against teams 1-40 than for wins and ties against teams 41-80. We also know that the bonuses are more for wins and ties in away games than in neutral site games, and in neutral site games than in home games. And, we know that the bonuses for wins (whether 1-40 or 41-80) are more than the bonuses for ties. We also know that a similar pattern applies for ties and losses against teams from 135 and lower in the rankings, with a break point at 205. In other words, the penalty groupings are 135-205 and 205-318. We also know, from basketball, that it is possible that the away-neutral-home amounts at one level of award may overlap the away-neutral-home amounts at the next level. For example, the awards for ties against good opponents could be 12-9-6 for away-neutral-home ties against teams 1-40 and 8-6-4 against teams 41-80). That's all we know.

What this means is that there are 12 possible bonus award amounts and 12 possible penalty amounts. A total of 24 variables. The only way to figure out the amounts is to make guesses, program those guesses into the system, and compare the rankings they produce to the NCAA's rankings. This is complicated by the possibility of games data inconsistency between the NCAA's data and the data of whatever person (me, in this case) is trying to "crack the code."

The NCAA staff has given me an informal explanation of why they keep these amounts secret. They cited two reasons: First, they don't want people attempting to clone the RPI and publish their results, because of the distinct possibility that what they publish will be wrong. (This happens regularly with basketball. I don't know if anyone has tried it in any other sport, beyond my trying it for women's soccer.) Second, they don't want people trying to make money selling their results to NCAA schools. You can form your own opinion as to whether either of these reasons makes sense.

I think there are other possible reasons why the NCAA doesn't release the amounts. One is that they don't want people really to understand how the complete system works and what its strengths and weaknesses are.

Another is that they don't want to put people in a position of being able to fully understand the information the Women's Soccer Committee had when it made at large selection and seeding decisions. This second one seems like a distinct possibility to me. Although the NCAA will be publishing the RPI three times during the season and once after the NCAA Tournament, the one RPI report it does not publish is for the RPI after completion of the regular season games (including the conference tournament games). Rather, it publishes an RPI report as of a week earlier (thus missing the last week's games) and the post-Tournament report (thus including the NCAA Tournament games). Of course, there is only one RPI report that matters, which is the one they don't publish after completion of the regular season games, because that is the report the Committee uses for at large selections and seeding. In fact, the RPI exists for the sole reason of producing that report. The gist of this reason for keeping part of the formula secret is to protect the Committee from criticism if others look at the data the Committee considered and the criteria and believe that the Committee made incorrect decisions. The NCAA may believe this protects the "integrity," at least in the public's mind, of the selection and seeding process.

My personal view on this is that if this is the NCAA's reason, then the NCAA is being "anti-educational" in a way that is unfitting for a representative of academic institutions. As a representative of educational institutions, the NCAA instead should release the full formula and the data the Committee considered and be happy to have people debate whether the Committee made the right decisions or not. I also believe that as a matter of sports promotion, that would be a better approach, as there are many sports fans who love these numbers games and would become stronger sports reporters if the numbers were out there in the open.

Keeping the pre-selection RPI secret came back to bite the NCAA a couple years ago when Oregon was frozen out of the tournament despite placing well in the Pac10 and beating USC and UCLA the last week of the season.

There were all sorts of accusations of politics. There was even the suggestion in a Eugene paper that Buzz Stroud, as chair of the West region, engineered the selections so we wouldn't have to face them.

I don't think they thought that through. The choice was face an unranked team team we beat 3-0 at in Eugene at home or go on the road at high altitude against dangerous #14 BYU and #10 Utah.... If Buzz engineered that he should have been fired.

The ducks just didn't make it because they had a low RPI O.C. schedule (Iona, PSU, Eastern Wash, New Hampshire) and a couple of bad losses (UC riverside and Long Beach St.)

UPSF has demonstrated once again just how valuable and insightful his amazing work on those dang RPI's is to Pilot Nation - and anyone else for that matter lucky enough to read his posts. If, at the end of the regular season, members of Pilot Nation are not the best educated and most RPI tuned-in posters in all of NCAA women's soccer land than it's our fault and not his. Great work UPSF and thanks for doing what you're doing!