Yet Another Sport That Gets it Wrong (Apparently)

On paper, it looks like there is reason to disagree over how the NCAA Basketball Tournament is seeded.

As a former wrestler, I have to admit, I do not “get” Hoopyjumpball (aka basketball in some social circles).  The less physical contact and combat in a sport, the less I understand it.  Therefore, I cannot and do not pretend to be a basketball expert.  ESPN pays people hundreds of thousands of dollars to blog about “Bracketology” all Basketball season, and for the most part, I’ll leave that to them.

However, as a God-fearing, red-blooded, all-American, male, I absolutely HAVE to pay my $5 and enter into one or more NCAA Basketball Tournament pool(s) every year.  Last year I started out trying to build my own predictive model similar to the models I have created for Baseball, and NCAA football.  Like all good modelers I first set out to understand who else and how else basketball performance was being forecast.  In that search, I stumbled upon this and so I stopped my own futile modeling quest.

When two professors from Georgia Tech come up with a predictive model that is apparently better than the NCAA’s for ranking teams, I know when it is time to quit.  The Linear Regression Markov Chain (or LRMC Model for short) apparently has had a ridiculous track record for picking games from the Sweet Sixteen forward.  As you can see in their report, from 2000-2006 they beat the pants off of RPI, Sagarin, Massey, AP, and ESPN.  Despite their best efforts, their tournament seeding system has not been adopted by the NCAA.  What a stubborn institution.  Between using the BCS and dismissing a methodology proven to be superior to others in existence it seems kinda silly.  On the other hand, if the tournament is seeded less scientifically, it at least adds more “fun” to the upset process.

Unfortunately, I tried to apply this model to my bracket last year and ended up being out another $5.  I’ve noticed that the site and its rankings have not been updated since 3/15/2009 either, suggesting that their bracket might have looked as bad as mine.

Last yer, the site provided rankings for every team in Division I.  Having made my picks simply on who was ranked “higher” didn’t pan out very well.  If Dr. Sokol and Dr. Nemhauser ever update the site again, The Game on Paper will do a Tournament break-down with the LRMC model.  If not, I’ll be forced to come up with some math to provide advice to my 5 readers during March Madness.  Happy New Year Mom!

Finally, a Sport That Gets it Right

On paper, Brent Metcalf and the Iowa Hawkeyes are #1 (again).

The week before Christmas through the week after New Year’s has to be the best three weeks of college sports during the year.  Of course, you have the the Football Bowl season beginning with the www.justsowecanhaveabowl.com Bowl and ending with the BS (I mean BCS) National Championship game.  Overlapping the Football season’s swan song is the end of the out-of-conference season for College Basketball and the start of the conference season.  Smack in the middle of this are also two of the biggest events in College Wrestling: The Midlands Tournament and the Cliff Keen/NWCA National Duals Championships.

The 2009 Midlands Tournament, which just ended last night, is one of Wrestling’s unofficial “Triple Crown” events.  Wrestlers who win their Conference Tournament, the Midlands Tournament, and the NCAA Tournament join the ranks of the truly elite in the sport.  In another week, the top College Wrestling teams in the country will be invited to compete in the Cliff Keen / NWCA National Duals Championships.  These two events are mid-season gut-checks for the wrestlers, but they also are a good litmus test for the institutions who rank the hundreds of student-athletes who compete in the world’s oldest sport.

What’s funny about wrestling is that there is not a lot of good data on the sport available to the public.  If you look at the NCAA’s “Statistics” page, you can find the goals and assists for every Women’s Field Hockey player in Division I – Division III, but you can’t find any wrestling statistics at all (and I thought Title IX was supposed to bring gender PARITY).  Despite this handicap, wrestling has one of the most intuitive (and more predictive) systems of pre-championship ranking available.

In College Wrestling there are 10 weight classes.  At the end of the season ~30 wrestlers from each weight class qualify for the NCAA tournament.  As wrestlers win matches they gain points for their team.  As you can imagine, wrestlers earn more team points by winning the championship and fewer team points the further away from 1st place that they end up.

Over the course of the season, most wrestlers compete in ~30-40 matches, and their records, who they beat, and how convincingly they won results in the NCAA tournament seeding.  Throughout the season, the ranking institutions (e.g. W.I.N. Magazine, USA Today, Intermat Wrestling, etc.) provide perspective on the rankings in each weight class based on past and future performance.  As a result, a wrestler’s total resume is taken into account for their individual ranking.

If we take the example of Brent Metcalf, the #1 ranked wrestler at 149-pounds, it is easy to understand why he is ranked #1.  Even if you have never seen Metcalf wrestle, the stats speak for themselves.  He has only lost 2 matches in college.  Both matches he lost were to Darrion Cladwell.  Caldwell is red-shirting this year.  Metcalf has faced and beaten the #2, #4, #6, and #13 ranked wrestlers at his weight class multiple times and in decisive fashion.  He was the National Champion in 2008 and the runner-up in 2009.  Wrestling is a fairly small world thanks to the elimination of numerous programs under Title IX, so this kind of breadth and depth of competition is fairly common in the sport.  This makes ranking the individual wrestlers easier over the course of 30+ matches where they are exposed to varying levels of competition and rated on their corresponding performance.

The team are then ranked on the sum of their parts using the very simple and intuitive TPI (Tournament Power Index) model sponsored by www.win-magazine.com.  As W.I.N. Magazine explains:

“The Tournament Power Index (TPI) is compiled by awarding points to each team for the ranked wrestlers listed in WIN’s current individual rankings. Teams are awarded points based on how many potential All-Americans they could have and get advancement points for wrestlers ranked No. 9-20. The order of teams in the TPI vary greatly from dual meet rankings, as some teams have a number of highly-ranked individuals but may have holes in their dual meet line-up.

Point totals associated with individual rankings are as followed: 1st – 20 (16 AA points + 4 advancement points); 2nd – 16 (12+4); 3rd – 13.5 (10+3.5); 4th – 12. 5 (9+3.5); 5th – 10 (7+3); 6th – 9 (6+3); 7th – 6.5 (4+2.5); 8th – 5.5 (3+2.5); 9th/12th – 2; 13th/16th – 1.5; 17th-20th – 1. TPI points do not match the final NCAA tournament points because it does not include bonus points for pins, technical falls, major decisions and forfeits/defaults.”

There is nothing nifty or fancy here, just doing the math behind the individual rankings.  If each of the rankings played out as predicted, that would be the score.  Basically it is a Condorcet ranking system that says the team with the best collection of wrestlers will win the tournament.

So, despite the lack of available data on wrestling, the team ranking system makes a lot of since.  And further, since its inception in the 2007-2008 season, it has correctly identified the University of Iowa to be the NCAA tournament champions.  Not surprisingly, the Hawks are #1 on paper again this year.

Alabama Spoils Florida’s Bid to Win the First-Ever TGOPoll National Championship. Bearcats Named Regular Season Champs

On paper, Tebow and the Gators Had Done It

In the last month, I had my own personal series of “Away Games” as a I hit Boston, Iowa City, Chicago, and New York City in a series of work and personal trips that kept me away from home and the computer for 2.5 of the last 4 weeks.

In that time, a lot happened in the world of NCAA football:  Ohio State won a coin-toss and will yet again represent the Big 10 in BCS Bowl, the Cincinnati Bearcats came back in baffling fashion to beat rival Pittsburgh, Greg McElroy continued his streak of undefeated games as a starting quarterback, the Texas Longhorns won again in-spite of Mack Brown,  and Notre Dame took my advice and showed Charlie Weiss the door, but then forgot to read my follow-up article and hired Brian Kelly.

In the mean time, The Game on Paper remained silent and watched as all these events unfolded.  The models kept crunching the numbers, but site remained silent.  This was unfortunate because it would have provided an opportuity to silence a lot of critics of the site. Over the course of the past three months, the poll has gotten a lot of flack for some of its rankings — notably Central Michigan.  However, the second biggest complaint I have received is that Florida was not ranked #1 despite ESPN, Bob Dole, and the American People all being convinced they were the best team in the country.

Well, in Week 13 of the season, the Florida Gators broke through, and it looked like all would be right with the world.  By the narrowest of margins, .001 points, Florida snuck past Cincinnati and based on the way things were shaping up for the games that weekend, it was conceivable that a blow-out win by Florida over Alabama and upset of Cincinnati by Pittsburgh would right the ship and set-up Texas and Florida to play in the National Title Game.

As we all know, that did not happen.  In fact, we almost saw Cincinnati in the National Championship game for real.

Alabama laid the spikes to Florida, and the Bearcats found 1 more point than the Panthers to take the Big East Title.  Now, the only thing that stood between a Cincinnati v. Alabama title game in both “on paper” and in reality was Texas v. Nebraska.

The Big XII Championship game turned out to be a barn-burner (thank God) instead of another 70-0 butt-kicking of some Big XII North Team.  In fact, Nebraska looked pretty damn good.  There is no question that Bo Pellini is a great coach, and I am still wondering what the hell the AD was thinking when he chose to hire Callahan after Nebraska canned Solich and won their Bowl Game under Pellini a few years back.  After the Nebraska v. Texas game, I was wondering why the hell the Texas AD sees fit to make Mack Brown the highest paid coach in the country.  The man benefits from coaching the flagship school in one of the largest, and MOST football-crazed states in the country.  Given those circumstances, honestly, I think I could coach Texas to a Bowl game.  Strangely enough ol’ Mack called a roll-out pass with :07 on the clock while in field goal range.  He had two time-outs.  He had great field position.  He had a quarterback who could not afford to get sacked (or injured) if they wanted to play in the National Title Game.  So he opted to throw the ball.  And the Nebraska fans went wild as it sailed out of bounds and time apparently expired.  Had McCoy put just a little more touch on the ball, they would have been right.  At least both Nebraska and Texas both found out what it felt like to both win and lose the Big XII Championship in the same day.

As a result, when the smoke cleared and the final numbers were crunched, Cincinnati and Texas

On paper, Cincinnati should be in the National Championship Game v. Texas

both vaulted over Florida to take the #1 and #2 spots for the season.  By the narrowest of margins, Florida remained the #3 ranked team despite a head-to-head loss v. Alabama.

This kind of “trick of the math” is among the thing most of my readers get most annoyed with.  However, as I have explained before, the objective of this poll is to assess all teams equally be the same measures.  Regardless of Head-to-Head games, what Conference they are from, or what the National Press says, the TGOPoll keeps to an objective standard of performance:

1.  Dominance of Competition
2.  Strength of Schedule
3.  VATI (Versus Average Team Index)
4.  Win/Loss Percentage

Further, although the TGOPoll rankings appear in a different order, they are not all that different from the final BCS rankings.  Of the final Top 25 teams in the BCS, 18/25 are ranked in the TGOPoll’s Top 25.  Of the remaining 7 team, Houston, East Carolina, Middle Tennessee State, and Central Michigan all appear in the Top 25 or “Others Receiving Votes” across ESPN, AP, Coaches, USA Today, and the Harris Polls.  Because the TGOPoll harbors no bias against Conferences like those polls do, it is not surprising those teams are given more love by the model than they are by subjective pollsters.

So as the regular season draws to a close, the TGOPoll is proud to announce that its selection for the Regular Season National Champion is the University of Cincinnati Bearcats.  I recognize this is little consolation in the wake of Kelly leaving and a double-overtime loss to rival Xavier in basketball, but at least someone out there respected the Bearcats this season.

Congratulations, and good luck against Florida.  On paper, you are better than the Gators.  Now, go prove it on the field.