From: "Steve Wrathell" To: Judiciary at mail dot house dot gov, president at whitehouse dot gov Cc: JSwofford at theacc dot org, JDelany at bigten dot org, KLW at big12sports dot com, MTranghese at bigeast dot org, BBBanowsky at C-USA dot org, RChryst at MAC-sports dot com, BGilliland at theMWC dot com, KWhite at UND dot com, THansen at pac10 dot org, MSlive at sec dot org, Waters at Sunbelt-sports dot org, KBenson at WAC dot org Subject: Bowl Championship Series, Congressional Hearings Date: Sat, 06 Sep 2003 01:17:47 +0000 Mr. Chairman and Committee members: I operate 2 computer football rankings systems, CPA Rankings and CPA Retrodiction Rankings, which rate teams for both the NCAA Division I (I-A and I-AA) and for the NFL. The former system utilizes victory margins and the latter does not and meets all stated requirements for the Bowl Championship Series. I began rating teams in 1997. Facts about the BCS. In every other sport, at top levels, there is a playoff to determine the champion, except in NCAA I-A football. For NCAA football there are 16-team playoffs for Divisions I-AA, II, and III. It should be noted that playing in a 16-team playoff final requires a team must play in 4 playoff games and that a few years ago, the University of Massachusetts (I-AA) complained that their playoff success was a substantial financial disaster for the school. I prefer playoffs, but the bows system has some advantages, too. No matter what the system is, it can be done in a way that equitably distributes funds or in a way in which teams feel cheated. In 1997, the year before the BCS, both Michigan (where I got my BA) and Nebraska went into the bowls as the only 2 undefeated I-A schools. They could not play each other because Big 10 and Pac 10 teams were committed to the Rose Bowl and the other major conferences were in the Alliance that allowed their top 2 teams to play each other. Michigan beat Washington State and Nebraska beat Tennessee and U-M and U-N were each voted #1 by one of the major polls, so the championship was split. The BCS solved this issue and would have had Michigan and Nebraska play each other had it existed in 1997. The BCS went beyond the old Alliance and the pre-Alliance bowl selection methods by creating the most fair system ever devised (which could still stand substantial improvement) for selecting teams for its bowls. Biased opinion polls alone would not be used. Polls would be combined with objective computer rankings and other objective criteria to determine the BCS Standings. Number 1 would play number 2. This is an improvement over anything of the past. Note that in 1984, BYU (now in the non-BCS Mountain West Conference) was the only undefeated team entering the bowls and was ranked #1, despite playing a weak schedule. It refused to play a highly-rated team for its bowl game, and chose to play Michigan, which was 6-5 going into that game. BYU won by just 7 points and was crowned National Champion. Had the BCS existed then, BYU would have been forced to defend its #1 status against the #2 team. In 1981, Clemson was undefeated and ranked #1, but played a weak schedule. Yet it chose to take on a highly-rated Nebraska team for its bowl. Clemson won the game and its #1 status was considered tested and legitimate. Of course, the "BCS teams" include members of 6 conferences, the ACC, Big 10, Big 12, Big East, Pac 10, SEC, and Notre Dame. Non-BCS teams are members of the Conf-USA, MAC, Mountain West, Sun Belt, WAC, and other I-A independents. I can assure you that the differences in the quality of the BCS teams, as a whole is well above that of the non-BCS teams as a whole. I can assure you that the quality of the BCS leagues' champions is above that of the non-BCS leagues' champions. However, when, last year, the Big 10 had 2 teams among the 8 in BCS games, their profits were shared with all 11 of the Big 10's members. That Ohio State and Iowa were better than Colorado State (MtnW), Marshall (MAC), Boise State (WAC), etc. is not in doubt. However the last few teams I mentioned were better than Michigan State, Indiana, and Northwestern, etc., which got nice slices of the BCS bowl money. Is that fair? Instead of focusing on only the 4 BCS bowls, one might want to also consider the other 24 bowls, too. There has been unfairness in who gets into those bowls, too. Non-BCS schools have complained that most of the bowls are already committed to having only BCS-league teams in their bowls. Until recently, there could be only one team from the MAC getting into any bowl and my computer ratings have shown the #2 MAC team being rated much higher than some teams that got into bowls when that #2 MAC team sat at home. Last season, South Florida, an independent, then, had a 9-2 record, beat teams that made it to bowls, but was denied admission to any bowl, while 7-5 teams with weaker schedule played in their place. South Florida is now in the Conference USA, so it now has a chance to get into a bowl. I believe that if there are to be 56 teams in the bowls, then there should first be an objective and fair determination as to which 56 teams deserve to be there and the bowls can select teams from that pool. Now, of course, there has been a requirement of having a winning record, which was relaxed to allow the weak Sunbelt Conference to get its champion into a bowl, but I believe this should be further modified to also use quality, objective computer rankings to additionally sort this out. I feel 2 "simple" things could be done to solve the BCS issues that would not affect the basics of the bowl system, and yet substantially change the economics of the bowls. 1) The 8 BCS bowl spots are not enough to accommodate the 11 conferences and the independents. The BCS should be expanded to include 4 more bowls, for a total of 8 bowls and 16 teams. There should be 4 spots reserved for champions of conferences that are not now in the BCS, as long as they finish in the top 25 of the BCS Standings. Added bowls could include the Cotton, Gator, and Capitol One Bowls, and one other. 2) Regardless of who is in the bowls, there should be profit sharing across Division I-A for ALL bowls (note that there are 28 bowl games in total, in I-A). I do feel, though that teams that play in the bowls should get a little more than non-bowl teams, to cover expenses. In the problem I mentioned above regarding U-Mass, their problem in playing 4 playoff games was that they incurred travel and other costs for those games, but split the revenues evenly with their conference members. Another issue with the BCS that should be discussed is TV. Are you concerned with monopolies? How's this?: of the 28 bowl games, this season and last, 25 are broadcast by ABC or ABC-owned stations (ESPN and ESPN2). One game each is broadcast by CBS, NBC, and Fox (the Sun, Gator, and Cotton Bowls, respectively). ABC controls ALL 4 of the BCS bowls, too! How much control does ABC have over the entire BCS process? I wish to note that government-run schools make up most of Division I-A schools. Usually, there are certain rules by which the state (and federal) governments live by, including disclosure and fairness. Yet, while the BCS standings cover 117 Division I-A teams, these standings are disclosed publicly for only the top 15 teams and the rest of the information is kept hidden. Errors in the BCS calculations and issues in the calculations had been discovered by Jerry Palm, who had published the 1-117 calculations as he determined them to be, under the BCS rules. Note that if the NCAA Division I-A were run like a professional sports league, the teams would probably be divided into several, relatively-evenly sized divisions of relatively even strength. Teams from the strong SEC would be blended with the weaker teams in that region from the CUSA and Sunbelt conferences. The 117 teams could be divided into 16 conferences (divisions) of 7 or 8 teams each. Toledo would be in the same conference as Ohio State, San Diego State would be in the same league as UCLA, etc. The rich would no longer be segregated from the poor. Schedule strengths would be much more comparable. And 16 divisions/conferences would be an ideal number for a playoff system. Note also that ALL of American college and pro sports, except for the NFL are having severe economic problems due to the struggle between teams that are in strong markets and those that are not. Bob Costas and George Will have testified before Congress about this, regarding baseball. The NFL has no such problems. All of its teams make money. Why? The NFL agrees with the Costas/Will concept: it takes 2 teams to play a game. NFL teams split the gate and TV revenues evenly. The home team gets 100% of only the concession and parking revenues. Thus New York teams and Seattle teams are comparable financially in the NFL and it is reflected on the field. No NFL team can now dominate year after year. The St. Louis Rams were 4-12 in 1998 and, in 1999, won the Super Bowl. Another issue, which will be a MAJOR issue if there is to be revenue sharing between the current BCS schools and the others in I-A is who should be in I-A? Some of the large schools are already trying to shrink I-A by increasing attendance requirements. About half of the MAC (Mid-America Conference) teams are at risk of being kicked down to I-AA. Meanwhile, I-A has been expanding. In 1998, there were 112 teams. In 2004, Florida A&M will play a I-A schedule, bringing the Division up to 118 teams. Yet, some people in the large schools would prefer a Division of around 80 schools. In 2003, Divisions I-A and I-AA are now roughly divided evenly. Counting Florida A&M as I-AA and including transitionals Cal-Davis and Northern Colorado (moving from Division II) as I-AA teams, there are 117 teams in I-A and 123 in I-AA. Another issue, which is closest to my heart is which computer systems are in the BCS. The first 3 systems in the BCS had major newspaper ties: the NY Times, Seattle Times, and Jeff Sagarin (USA Today) systems. Then, the computer systems were expanded to eight, and a stated criterion was that the systems had to be in operation for at least 5 years (I was just short of that at that time). Later, there was political backlash against computer systems that used victory margins, so some changes were made, and some of the involved systems changed or their creators established an additional system that did not use victory margins. Then, all victory margin systems were banished and the systems of Herman Matthews, Dunkel, and the NY Times were removed and were replaced. Later, the NY Times was brought back after it had an untested non-victory-margin system. Whether or not one agrees with victory margins in computer rankings, the question remains: which systems should be used. I have published studies on the accuracy of computer rankings and John Wobus has continued with a methodology similar to mine, but with an even broader scope. One will note that the NY Times' system always rates at the bottom. Those measurements rated predictive accuracy, which disfavors non-victory-margin systems. The most accurate system for measuring systems on non-victory-margin accuracy is maintained by Kenneth Massey (who has a system in the BCS). Each week, he calculates each systems' "Ranking Violation %." If a system had a Zero rating, then in no instance does it rate a team below a team it beat. It is impossible to be at zero at the end of the season, since bad teams do beat better teams. But the best non-victory-margin systems can be measured this way (and victory-margin systems can also be measured this way too.) The following was on my web site at the end of last season: ----------------------------- RANKING VIOLATIONS - BCS & CPA SYSTEMS COMPARED CPA Retrodiction and BCS Systems' Ranking Violations (As determined by Kenneth Massey. See the CF Rkg Comparisons, linked at my description page.) Massey's codes are used. 108 means 10.8% (etc.) of the time, a system ranks one team ahead of a team that beat it. Lower numbers are better. This is a retrodictive, not a predictive, measurement. Division I-A: (recent results by weeks ended...) Sys N 09 N 16 N 23 N 30 D 07 Final Avg. CPR 99 138 135 146 144 126 131.3 WOL 127 146 157 149 144 156 146.5 MB 129 141 155 154 150 164 148.8 SE 134 138 160 154 153 160 149.8 COL 149 150 154 149 149 167 153.0 AND 144 151 154 158 152 171 155.0 BIL 138 155 180 181 180 165 166.5 NYT 149 167 172 193 187 198 177.7 Division I-AA: (recent results by weeks ended...) Sys N 09 N 16 N 23 N 30 D 07 Final Avg. CPR 134 172 153 174 182 159 162.3 SE 151 165 175 189 186 184 175.0 MB 165 181 180 187 184 186 180.5 WOL 167 174 186 187 189 184 181.2 Of 81 I-A systems on Massey's Comparison, 76 listed all 117 teams. Of those, NYT beat 7 systems (mostly or all predictive) and 73 systems beat NYT for ranking violation. The worst and third worst RV's were 340 and 271, by USA and AP (the polls), respectively. RANKING VIOLATIONS - THE BEST SYSTEMS FOR 2002 Division I-A, Final (Post-bowl) Top 5 - All Systems Rnk RV Code System Name 1 126 CPR CPA RETRODICTION RANKINGS 2 132 MCK McCormick / GACFF 3 141 CPA CPA RANKINGS 4 156 WOL Wolfe (BCS) 5 157 BIH Bihl Rankings Division I-AA, Final (Post-playoffs) Top 5 - All Systems Rnk RV Code System Name 1 159 CPR CPA RETRODICTION RANKINGS 2 176 WIL Wilson Performance 3T 184 SE Sagarin-Elo (BCS) 3T 184 WOL Wolfe (BCS) 3T 184 CSL CSL Ratings (Loest) Note: I'm not sure, but I think this is the first time since Kenneth Massey has published the RV's that the same system has been #1 for both I-A and I-AA. CPA Retrodiction: #1 in I-A and I-AA for R.V.! --------------------------------- Massey's Comparison is at: http://www.mratings.com/cf/compare.htm http://www.mratings.com/cf/compare1aa.htm If the computer systems were chosen for the BCS based on quality alone, one of my systems would be among them and the NY Times and Seattle Times systems would not. The selection process is made behind closed doors in secret meetings without objective criteria. This isn't how it should be done in an organization made up of primarily state-owned colleges. It should also be noted that while the BCS bowls generate tens of millions of dollars, the operators of the computer systems that help determine the teams get paid ZERO, NOTHING, ZIP, NADA, for their efforts. Yet the BCS pays another entity $100,000 per year to do some of the computations for the BCS standings. Some of the computer raters get paid by newspapers, but some do it for absolutely no economic benefit. Is this right? Should objective computer rankings be ignored and should bowl selection be based on subjective polls or by small secret committees? Read my essay: "Why Computer Rankings are Better than Polls http://www.cae.wisc.edu/~dwilson/rsfc/rate/wrathell3.txt Do I know anything about football rankings? Take a look at this info from my web site: AWARDS AND ACCOMPLISHMENTS: The CPA Rankings computer football rating system won several of Todd Beck's Prediction Tracker Awards (and honorable mentions) in several categories, predictive and retrodictive, for both the NCAA I-A and the NFL. 2002 NCAA IA - WINNER!!! - BEST OVERALL RETRODICTIVE SYSTEM 2002 NCAA IA - 2nd Place - Most Retrodictive Wins 2002 NCAA IA - 2nd Place - Best Retrodictive Mean Absolute Error 2002 NCAA IA - WINNER!!! - Best Retrodictive Mean Square Error 2002 NCAA IA - WINNER!!! - Smallest Retrodictive Bias 2002 N F L - Co-Winner - Best Straight Up Winners, Second Half 2002 N F L - 2nd Place - Smallest Bias, Second Half 2002 N F L - 2nd Place - Best Overall Retrodictive System 2001 NCAA IA - Hon.Mentn.- Smallest Bias, Entire Season 2001 NCAA IA - WINNER!!! - BEST OVERALL RETRODICTIVE SYSTEM 2001 NCAA IA - WINNER!!! - Most Retrodictive Wins 2001 NCAA IA - WINNER!!! - Best Retrodictive Mean Square Error 2001 NCAA IA - WINNER!!! - Smallest Retrodictive Bias 2001 N F L - WINNER!!! - BEST OVERALL PREDICTIVE SYSTEM 2001 N F L - WINNER!!! - Best Against the Spread, Entire Season 2001 N F L - WINNER!!! - BEST OVERALL RETRODICTIVE SYSTEM 2001 N F L - 2nd Place - Best Retrodictive Mean Absolute Error 2001 N F L - 2nd Place - Best Retrodictive Mean Square Error 2000 NCAA IA - Co-Winner - BEST OVERALL SYSTEM (Pred. & Retrod.) 2000 NCAA IA - 2nd Place - Best Mean Absolute Error, Entire Season 2000 NCAA IA - 2nd Place - Best Mean Absolute Error, Second Half 2000 NCAA IA - WINNER!!! - BEST OVERALL RETRODICTIVE SYSTEM 2000 NCAA IA - WINNER!!! - Most Retrodictive Wins 2000 NCAA IA - WINNER!!! - Best Retrodictive Mean Absolute Error 2000 NCAA IA - WINNER!!! - Best Retrodictive Mean Square Error 2000 NCAA IA - WINNER!!! - Smallest Retrodictive Bias 2000 N F L - WINNER!!! - Best Straight Up Winners, Entire Season 2000 N F L - WINNER!!! - BEST OVERALL RETRODICTIVE SYSTEM 2000 N F L - WINNER!!! - Most Retrodictive Wins Note that CPA Retrodiction had the highest win % in the NFL in 2001, but did not win a P.T. Award, since it began at mid-season. No other system has won as many P.T. Awards. http://tbeck.freeshell.org/"> PRESEASON ACCURACY: Chris Stassen's calculations of the most accurate web sites and magazines for NCAA conference prediction accuracy is online. Results: OK, so I was only 3rd for 2002, but for the last 3 years: 3-yr Magazine / Web Site 3-yr 1 CPA Rankings ------ 420 CPA Rkgs was #1 in 2000. 2 Lindy's ----------- 429 3 Phil Steele's ----- 431 Steele was #1 in 2001. 4T Jim Feist's ------- 439 4T The Sporting News - 439 TSN was #1 in 2002. 6 Athlon ------------ 440 Lower numbers are better. (c) 2003 Chris Stassen For the complete chart and much more info, to go: http://preseason.stassen.com/prediction-accuracy/2003-3yr.html CPA Rankings is one of the components of the Gridiron Power Index (GPI), which is a BCS-style ranking method for I-AA teams. http://i-aa.org/section_front.asp?arttypeid=564 Thank you for reading thru my information. May God grant you wisdom in the handling of these and other matters. Sincerely, Steven Wrathell, CPA President, Steven Wrathell, CPA, PC CPA Rankings / CPA Retrodiction Rankings http://www.cae.wisc.edu/~dwilson/rsfc/rate/wrathell.html BA - University of Michigan, Ann Arbor, MI, 1982 MSPA - Walsh College, Troy, MI, 1985 P.S. The US Constitution can be amended only thru the Constitutional amendment procedures and not by the Supreme Court or by any judge, contrary to current practice. Hasn't anyone ever read the Federalist Papers, etc.? Thomas Jefferson was greatly upset by the Marbury v Madison decision. How do you think he and James Madison would react to some of the recent Supreme Court decisions? When is Congress going to do something to defend the WRITTEN Constitution?