2011 rpi

Massey's rating system now is up and running for the 2011 season. Use the following link: http://www.masseyratings.com/rate.php?lg=csocw

At the Massey webpage, you can select the ratings for whichever college level you want to see (e.g., Division I).

For unofficial, but reliable, Division I women's RPI ratings, use the following link to the nc-soccer pages: http://www.nc-soccer.com/wsoccer/2011/

Please note that Massey's system starts with an initial "weighting" of teams based on past history. Over the course of the season, Massey winnows out the weighting as he inputs more game result data into the system. By having an initial weighting, Massey makes his ratings currently usable for considering the relative strengths of teams.

The RPI, on the other hand, starts out with the assumption that all teams are of equal strength and moves away from that assumption based only on games played. It is not intended to be usable currently for considering the relative performance of teams, due to the very limited current data within the system. The only time it is intended to be usable for considering how teams have performed is at the end of the regular season, when it is used in the NCAA Tournament at large selection and seeding processes.

 

I have posted, at the "RPI for Division I Women's Soccer" website, an RPI report for games through Monday, September 5. The report is in the form of an attached downloadable Excel workbook, with three spreadsheet pages. The first page covers teams' RPIs, the second conferences' average RPIs, and the third regions' average RPIs. Use the following link to get to the "RPI Reports" page: https://sites.google.com/site/rpifordivisioniwomenssoccer/rpi-reports. Given it is very early in the season, the RPI ratings right now mean little in terms of what teams' ratings will be at the end of the season. Rather, the current ratings mostly are useful for learning how the RPI works and develops over the course of a season.

As an added item this year, the reports include ratings using a "Modified RPI." The Modified RPI does three things:

1. It uses a simplified calculation of Element 2 of the RPI. If I'm calculating Team A's RPI rating, then the NCAA's Element 2 is the average of Team A's opponents' winning percentages against teams other than Team A. A result of the "against teams other than Team A" qualification is that Team B can contribute one winning percentage to Team A, another to Team C, and another Team D, etc., notwithstanding that Teams A, C, and D all played the same Team B. The simplification used in the modified RPI eliminates the qualification so that Team B's contribution is the same for all its opponents.

2. It takes home/away imbalances into account. It does this by giving a team a -0.008 adjustment for each home game and a +0.008 adjustment for each away game and averaging out these adjustments over all games to come up with an overall game location adjustment for each team.

3. It makes regional adjustments based on the average RPIs of teams within each of five regions. This is to address the RPI's problem rating teams from different regional playing pools in a single system, which results in the NCAA RPI's discriminating against teams from strong regional pools and in favor of teams from weak regional pools.​

Information on the rationale and research for, and details of, each of these adjustments is covered on the RPI for Division I Women's Soccer website. When you're at the website, check out the "Element 2 Issues," "Home/Away/Neutral Issues," and "Regional Issues" pages, respectively, and then the "Modified RPI?" page. I've made major revisions to the website in order to incorporate my last year's research, and these all are new pages.

Although the data for the four years 2007-2010 indicate that the Modified RPI would be an improvement over the NCAA RPI, the modifications do not make a great difference although they make some difference. I'll be including the Modified RPI ratings in my reports over the course of the season, to see and show how the modifications would affect the current season, if at all, when it gets to NCAA Tournament bracket formation time.

One further note: For those who are interested in knowing what teams RPI ratings are at any particular time, I recommend the NC-Soccer website here: http://www.nc-soccer.com/wsoccer/2011/index_arpi. You can navigate around that website to get lots of information. At this point, the RPI for Division I Women's Soccer website is more for people who want to know the intricacies of how the RPI works, the details of how well it does at measuring teams' performance, its weaknesses, and so on. Also, with the RPI reports being downloadable, it allows anyone interested in studying the weekly development of the RPI to do it once the season is over.

 

I have posted a new RPI Report on the RPI for Division I Women's Soccer website, covering games through Sunday, September 11. The report is in the form of an Excel file attachment at the bottom of the website's RPI Reports page. Those interested can either open the report and use it on line or download it and use it at your leisure. The file title is 9.11.2011 RPI Report. Here is the link to the RPI Reports page:

https://sites.google.com/site/rpifordivisioniwomenssoccer/rpi-reports

 

NCAA Game by Game Statistics

The NCAA's Game by Game statistics system is up and running. Today is the first day as of which schools are supposed to have reported all of their game results and stats to the NCAA, for games through yesterday, September 11. The system covers both game results and game statistics, including individual player statistics.

The basic reporting requirement from here on out is that schools report their game results and stats not later than noon Eastern Time each Monday, covering games through the preceding day (Sunday). Overall, the system is working well and is very nifty. Here's a link that will get you into the system: http://stats.ncaa.org/team/inst_team_list?academic_year=2012&conf_id=-1&division=1&sport_code=WSO

School Information Directors (SIDs) are responsible for reporting their home game results to the NCAA. When two teams play in a tournament at a neutral site, those two teams are responsible for deciding between themselves which SID will report the results (the host school cannot report the results, as the system will accept results only from the two teams that played the game). Although the home team's SID is responsible for reporting the results, the away team's SID also can report the results; and both teams' SIDs can report results for neutral site games.

There also is a procedure for making corrections, if there are errors in what is reported. And, there are procedures for changing scheduled game dates and locations.

I believe the NCAA sends out reminders if there are games on the schedule for which it has received no reports. They definitely do this if there is no report for a neutral site game.

As of right now, there are only 10 games that have not yet been reported into the system. They are the following, with the home team listed first:

8/19 Pacific v Nevada
9/8 Northern Illinois v Wisconsin
9/9 Mississippi Valley v Louisiana Lafayette
9/9 Pacific v UNLV
9/9 Georgetown v James Madison
9/11 Mississippi Valley v SE Louisiana
9/11 Iowa v South Dakota
9/11 St Peters v Monmouth
9/11 Georgetown v St Francis
9/11 Ohio v Ohio State

Check out the NCAA's system. It's a great source for team and individual player stats.

 

I have posted a new RPI Report at the RPI for Division I Women's Soccer website, covering games through Sunday, September 18. It is in the form of an Excel spreadsheet attachment at the bottom of the website's RPI Reports page. Use the following link to go to that page: https://sites.google.com/site/rpifordivisioniwomenssoccer/rpi-reports

The report file title is RPI Report 9.18.2011.

Page 1 of the report shows details of teams' RPIs and strength of schedule, as well as what their RPIs would be under a modified RPI that takes game locations and regional strength into account.

Page 2 of the report shows conferences' average RPIs. It confirms what people have been supposing, which is that the ACC is having a very strong year. It also shows the SEC and Big Twelve as having good years. It also shows the West Coast Conference as having a good year. The Pac 12 is off from previous years. Also, the Big Ten and Big East are not having particularly good years.

Page 3 of the report shows regional playing pools' average RPIs. It shows the Southeast pool as being the strongest, followed by the West. This is unusual, as for the last four years the West has been the strongest.

 

I'll post this here, since it relates to the NCAA Tournament and I don't want to start a Tournament thread this early in the season. The NCAA just has released its 2011 Pre-Championship Manual. It contains a lot of information about the Tournament. I haven't read the whole thing carefully yet, but there are some significant changes from past years. We already know that there's a change so that the first weekend will have only one game and the second will have two games. But, the changes go beyond that. Here's a link for those who are interested. Check it out: http://ncaa.org/wps/wcm/connect/af8...ERES&CACHEID=af815880486989b490ccb74c8af46e6f

 

Regarding the NCAA Tournament, bmoline you're probably right that there are not a lot of significant changes. I've had a chance to look more carefully at the NCAA's Pre-Championship Manual and here's what I see:

1. As we already know, the format this year will be:

First weekend, Round 1

Second weekend, Rounds 2 and 3 at a common site

Third weekend, Quarterfinals

Fourth weekend, College Cup in Kennesaw, GA​

There is an important new paragraph in the Manual, on page 7 within the "General Requirements" section. Here is what it says:

"For non-revenue generating championships, that seed a quarter of the bracket, sites will be awarded to the seeded teams who submit a bid that meets the requirements for hosting. In the event that a seeded team does not submit a bid that meets the requirements, the sports committee will maintain the current policy of geographic consideration when assigning those host sites. The committees will use the geographic considerations to help maintain the balance and integrity of the bracket. The committees will also separate first and second round conference opponents when assigning teams to sites.​

The way this is written, it appears to represent an NCAA policy that now applies to all non-revenue generating championships, which at this point includes Division I Women's Soccer. Although I don't have a lot of confidence in the overall draftsmanship for the Pre-Championship Manual, I'm going to interpret the above language as an attorney would, given that it comes from an overall NCAA policy and is recently written:

1. As seems obvious, the seeded teams will host their first round games, so long as they have a qualifying field and submit an appropriate bid to host. Although the Manual is silent on who will host the first round games that do not involve seeded teams, I'm guessing it will be the team with the better RPI if it has a qualifying field and has submitted an appropriate bid to host. Alternatively, it could be the team that submits the better bid, if the NCAA finally is going to look at this from an economic perspective.

2. More importantly, a seeded team will host the second/third round games so long as one of the two teams has a qualifying field and has submitted an appropriate bid to host. ("ites will be awarded to the seeded teams who submit a bid that meets the requirements for hosting.") This is true regardless of geographical considerations. If neither seeded team qualifies to host, then the site will be decided based on geographical considerations. This suggests to me that the NCAA will need to place non-seeded teams in the bracket, using geographical considerations, based on the assumption that one of the two seeded teams will be hosting. It may suggest that the NCAA, in placing seeded teams from the four seed pods (see below) in relation to each other, will place seeded teams from the same geographic area within the same 8-team bracket groups, to the extent possible. This would make the first three rounds something like regional qualifiers.

3. If both seeded teams playing in a second/third round foursome have qualifying fields and submit appropriate bids, the Manual does not say which team will get to host. Here too, it could be the higher seeded of the two teams or it could be the one that submits the better bid.

4. If a seeded team and an unseeded team play in the quarter-finals, the seeded team will get to host if it has a qualifying field and has submitted an appropriate bid. If two seeded teams play in the quarter-finals, then it again isn't clear whether the higher seed will get to host or the one that submits the better bid.

5. The previous few years' policy, that teams from the same conference cannot meet in the first or second rounds, will continue. This doesn't mean that two teams from the same conference can't be at the same second/third round site, but it does mean that they won't meet in the second round game.

6. Subject to meeting all of the above requirements, the placement of teams in the bracket will be based on geographic considerations "to help maintain the balance and integrity of the bracket," (whatever the quoted language means).​

There is some other language that is a carryover from the past and part of which is wrong. This is on page 12 under "Pairings":

The top 16 teams identified by the committee will be seeded in the bracket. The seeded teams will compete at separate first and second round sites. [This clearly is wrong, since the bracket is set up so that there will be two seeded teams at each second/third round site.] The committee has been given approval by the Championships/Sports Management Cabinet to place the top 16 seeds in pods of four. There will be four number ones, four number twos, four number threes, and four number fours. The remaining 48 teams will be paired geographically. First- and second-round matchups will be avoided. The committee will use the NCAA's mileage calculator located on the travel Web site at http://web1.ncaa.org/TES/exec/login?js=true when establishing the mileage for travel. Teams within 400 miles of each other will be required to drive.​

Beyond this, I did not see any significant changes in the Manual. (It continues to slightly mis-describe the bonus/penalty system for going from the RPI to the Adjusted RPI.)

 

Here are two other significant pieces of information from the NCAA's Pre-Championship Manual:

1. According to the Manual, the NCAA will release weekly RPI reports on NCAA.com, beginning September 20 and ending November 1. If true, this will be the most RPI reports the NCAA ever has released, and the earliest initial release date. For whatever reason, however, the September 20 RPI report still is not posted on NCAA.com, so far as I can tell.

2. The members of the Women's Soccer Committee are identified on page 8 of the Manual. In addition, the members of the eight Regional Advisory Committees are identified in Appendix G, at pages 34-35.

 

I have learned that the first NCAA RPI rankings for the 2011 season will be released next week.

For those of you interested in:

The detailed rules that will apply to the at large selections, seeding, and bracket formation for the NCAA Tournament and

The NCAA Division I Women's Soccer Committee's procedural steps involved in getting from now through completion of the bracket​

I've updated the pages on those topics using the most current information available, on the "RPI for Division I Women's Soccer" website. The following link will take you to the "NCAA Selection, Seeding, and Bracketing Criteria" page and you can go from there to the "NCAA Bracket Procedure" page.

The link: https://sites.google.com/site/rpifo...caa-selection-seeding-and-bracketing-criteria

 

I have posted a new RPI Report at the RPI for Division I Women's Soccer website, covering games through Sunday, September 25. It is in the form of an Excel spreadsheet attachment at the bottom of the website's RPI Reports page. Use the following link to go to that page: https://sites.google.com/site/rpifordivisioniwomenssoccer/rpi-reports

The report file title is RPI Report 9.25.2011.

 

The NCAA how has published its first RPI rankings for the season, "Last updated - September 28, 2011." I don't know yet what the cutoff date was for games included in the report. Here's a link: http://www.ncaa.com/rankings/soccer-women/d1/ncaa_womens_soccer_rpi

EDIT: Based on a quick check of teams that played games on Monday and Tuesday, it looks to me like the last date for which games were included in the report is Monday, September 26, except that the DePaul v South Florida game played on Monday is not included because no one has reported the game stats to the NCAA yet.

 

In reviewing the NCAA's just published first RPI report for the season, I found that they are including the results of Nebraska-Omaha in their calculations. Nebraska-Omaha is in the first year (the "exploratory year") of its transition into Division I. Here is what the NCAA's 2010-2011 Statistics Policies have to say about consideration, in the RPI, of the results of teams in their first year of transition into a division:

Reclassifying Teams in Rating Percentage Index (RPI) and Statistics
An institution that is changing divisions goes through a multi-year process to enter another division. The process is the same whether the school is moving up or down in divisions. The first year of this process is called an “exploratory year”. If moving to Division I, the second year after the exploratory year the institution must be playing a Division I schedule and will be included in the RPI. An institution that arranges a Division I schedule the year following its exploratory year could be included in the RPI if it notifies the NCAA by September 15.

A school in its exploratory year will be included in the weekly statistical rankings in its old division. If the school continues with the process, that reclassifying school will be included at the bottom of the weekly statistical rankings once the scheduling requirements for its new division have been met. It is required to do this by its second year after the exploratory year. In the first year after the exploratory year, if a school moving to Division I has met the criteria to be included in the RPI, it will also be included in the statistical rankings. These reclassifying schools will not be ranked but rather listed at the bottom of the weekly rankings in all individual and team categories for which they would qualify. However, if a school changing divisions has not met the scheduling criteria in the first year after the exploratory year, it will be included in the statistics for the division it is leaving. Once a reclassifying school has reached full-fledged membership in its new division, it will be ranked along with the other schools in that division.

....​

(Emphasis added.)

I've looked briefly through the Statistics Policies, and I don't find any provision for waivers of the policies. Perhaps a waiver provision is buried there somewhere or, alternatively, perhaps there is a policy of granting waivers even if there is not explicit authority for waivers. In any event, however, I asked the NCAA staff whether the inclusion of Nebraska-Omaha's games in the RPI calculations was a mistake and they have advised me that Nebraska-Omaha asked for, and was granted, a waiver. Unfortunately, although Nebraska-Omaha may have been aware of this, I'm not aware of the NCAA ever having advised anyone else of it, notwithstanding that it could have an effect ultimately on the NCAA Tournament bracket formation process.

From a practical perspective, I think it is rather odd that the games of teams in their first year of transition ordinarily are not included in the RPI calculations. On the other hand, I think the NCAA should announce to all a divisions' teams if it is going to issue a waiver. In addition, if there indeed is authority for waivers, it should be so stated in the Statistics Policies.

In any event, both nc-soccer and I now get to revise our systems to include the Nebraska-Omaha games.

 

Part of the "not fun" of working to produce absolutely correct RPI ratings begins once the NCAA starts publishing its RPI reports. It consists of matching their win-loss-tie and away-home-neutral totals for teams with mine and then, if there are inconsistencies, figuring out why. It was during this process that I discovered that the NCAA is including UN Omaha's game results in its RPI calculations. (Except for that, my and nc-soccer's win-loss-tie totals match the NCAA's perfectly.) I discovered a few game location errors in my data -- and one in nc-soccer's -- that have been corrected. I also discovered eight game location errors in the NCAA's data, of which I've advised them and that I expect will get corrected. Game location errors affect only the bonus/penalty adjustments, so they're not a huge deal although, under the right circumstance, they could make some difference say for a team at one of the edges of the NCAA tournament at large selection "bubble." The particular errors I found are instructive on how game site errors find their way into the system.

Three errors occurred when games showed up in the NCAA's database as "neutral site" games. In each case, the data enterer input a geographic location as the game site, as follows:

Charlotte v South Carolina entered as at Columbia, SC
Drexel v Delaware entered as at Philadelphia, PA
Davidson v Florida Gulf Coast entered as at Davidson, NC

These three games indeed were at those cities, but they were at South Carolina, Drexel, and Davidson. The problem is, when you enter a geographic location for a game site, the NCAA's program considers that a neutral site game. To do it properly for a game that is not at a neutral site, you have to enter the school as the site. Thus although the above three games were at South Carolina, Drexel, and Davidson respectively, they are identified by the system as at neutral sites.

Five errors occurred when neutral site games showed up as @ one of the two teams playing. This is the error I've seen the most over the past few years. What happens is this: for purposes of a tournament a school is holding, when two schools from elsewhere are playing each other, one is identified as the home team and one as the away team. Really, neither is home and both are at neutral sites. The person entering the data into the NCAA system, however, sees one team identified as home and the other as away and enters it into the system that way, rather than realizing that both are playing at a neutral site. The NCAA's guidelines for entering data into the system warn against this error, but hey ....

 

I discovered something interesting that is a little disconcerting:

The NCAA's Game by Game stats are available here: http://stats.ncaa.org/team/inst_team_list?academic_year=2012&conf_id=-1&division=1&sport_code=WSO.

Check out Oregon by clicking on "Oregon." Count up the number of away-neutral-home games. As of right now, it is 2-5-3. In counting, remember that the game @ Eugene, OR is in the data base as a neutral site game, not as a game at the University of Oregon. Then go to the NCAA's current RPI report available here: http://www.ncaa.com/rankings/soccer-women/d1/ncaa_womens_soccer_rpi. Count up the number of away-neutral-home games. As of right now, it is 2-4-4.

The difference between the two is that the game listed in the Game by Game stats indicates the Oregon v UNLV game was at a neutral site ("@ Eugene, OR"), whereas the RPI report properly includes the game as at the University of Oregon.

The RPI, ordinarily, draws its data from the team data reports that result in the Game by Game pages. Apparently, however, the NCAA RPI staff can intervene and make corrections (as they apparently did for Oregon after being advised that the Game by Game stats had an erroneous neutral site listing whereas the game was at the U of O). What's disconcerting, however, is that the Game by Game stats still contain the error. One would hope that the Game by Game errors would be corrected at the same time that the data drawn from the Game by Game stats are corrected.

 

Regarding the immediately previous post, if anyone is interested, here's my understanding as of today on how the NCAA's stats input system works:

The schools input their schedules into the system in August. At some point in August, the NCAA's master schedule is "locked" and after that, if schools want to add games or make schedule revisions, they go through the NCAA stats staff person responsible for women's soccer.

As games are played, the schools prepare xml files (whatever they are) and send (email?) them to the NCAA. These files consist of the game stats, including results. The NCAA staff inputs these files into its overall "official" statistics system. The staff also extracts the appropriate data from these files and inputs it into a separate RPI system. I don't know what the "input" process is, but I assume it occurs electronically rather than a staff person having to read the xml file and then manually enter the data into the systems.

The NCAA staff cannot change the data entered into the "official" stats system. Only the schools involved in a particular game can do that. Thus if the staff knows of an official stats error, or anyone else does, the only way to get it corrected is to notify the schools involved in the game and hope they correct it. On the other hand, the NCAA staff responsible for the RPI fir a particular sport can change the data in the RPI system, when they find out about an error. I was right, in the previous post, that this was how the RPI data ended up with the correct site for the Oregon v UNLV game even though the "official" Game by Game stats have the incorrect site.

The problem for the NCAA, with getting "official" stats errors corrected, is that they have a massive number of xml files coming in that need to be translated into the "official" stats system (~5,000 soccer files so far this year, which I believe includes all NCAA soccer not just Division I). That process is challenging, which explains why going through the process of getting schools to make game location corrections to the official stats may not be occurring even though the corrections are being made to the RPI data.

I would rather it were otherwise so that the two systems would match, but I do understand the practical problem the NCAA has.

 

CW, what you describe is what I assumed, but didn't know, was the case, so thanks for your great info. I think we now have a pretty good picture of how the system works. I don't think that the system is set up to enter a neutral site as a default. Rather, I think sometimes the home school, which is supposed to transmit the XML file, sends something like site>Eugene, OR<site, to use the University of Oregon as an example, rather than site>University of Oregon<site. The NCAA software then treats the former as a neutral site whereas it would treat the latter as a game at the U of O.

This all might be of not much interest to others, but I really try to understand what the difficulties are for the NCAA staff so that I'm not overly critical when errors show up.

 

10/2/2011 RPI Report

I have posted a new RPI Report at the RPI for Division I Women's Soccer website, covering games through Sunday, October 2. It is in the form of an Excel spreadsheet attachment (downloadable) at the bottom of the website's RPI Reports page. Use the following link to go to that page: https://sites.google.com/site/rpifordivisioniwomenssoccer/rpi-reports


The report file title is 10.2.2011 RPI Report.

Page 1 of the report shows details of teams' RPIs and strength of schedule, as well as what their RPIs would be under a modified RPI that takes game locations and regional strength into account.

Page 2 of the report shows conferences' average RPIs.

Page 3 of the report shows regional playing pools' average RPIs.

 

I have a question for those of you out there who are D I school-affiliated. Do you have access to the NCAA's weekly RPI reports other than through the posting we all can see on NCAA.com?