Friday, August 17, 2007

the Gunther Rating System: How it Works

With these computer rankings, I've tried to create a system based on logical thinking that analyzes a few easily recognizable traits of every college football game. Since the goal is to objectively separate, rate, and rank college football teams throughout the course of the season, in my opinion, the most successful teams:

1. Win a lot of games, and win against strong competition
2. Win on the road (and don’t lose at home)
3. Win by large margins of victory (or keep it close if they do lose)
4. Win at the end of the season

Teams earn points within these four categories, and a team's point total is the sum of all four plus a portion of their conference average.

1. Who (Winning + SoS)
Principle: Who Points are earned/lost depending on who the opponent is.
Explanation: In a sport where nothing is as important as the W, this category rewards winning and penalizes losing. The best teams will have very few (if any) losses at the end of the year. But this category also takes into account schedule strength and will separate further the teams who win against good opponents from the teams who win against bad opponents. A win against a great team (who has a lot of points) will earn maximum points, while a loss to a bad team (who has negative points) will lose maximum points.

2. Where (Home-field Advantage)
Principle: Where Points are earned for an away win and lost for a home loss.
Explanation: Playing at home is an advantage, with home teams winning approximately 60-65% of all games depending on the year. The best teams win no matter where they play, overcoming home-field advantage when they play away and using it when they play at home. The yearly home-field advantage percentage and opponent strength is used to determine how many points teams earn or lose per game.

3. How (A for Effort)
Principle: How Points are earned for a large Margin of Victory win or small MoV loss, and they're multiplied by the opponent's strength.
Explanation: There's two unique things about my MoV component. The first is that it is tallied by possession, not straight points. For instance, a MoV of 4 earns the same as a MoV of 8, since both mean that the opponent could have won or tied in one possession. Over 8 would take two possessions, over 16 three. The more possessions it would take your opponent to win or tie, the higher your How Points earned. The second unique thing is that the losing team can earn How Points if they kept the score close. Losing by 3 points or less earns maximum How Points, between 4-8 pts earns some How Points, etc.

There’s two sportsmanship fail safes with this MoV system. First, it’s capped at 17, or three possessions, making it disadvantageous to run up the score any higher than that. Second, by giving points to the losing team for keeping it close, the winning team will actually earn a few more Who Points because their opponent will have be stronger. It’s not so many that it would be worth letting the opponent get the score closer (and jeopardizing the win), but it would be enough to influence the decision of either taking a knee or punching it in on 2nd & Goal at the end of the game.

4. When (Progress)
Principle: When Points are lost depending on how far into the season a team’s 1st and 2nd losses occur.
Explanation: This category is designed to separate the teams at the very top of the ratings. I believe that the best teams get better as the year goes on, developing and progressing the more games they play. Losing early in the season is somewhat understandable, but losing in November, even to a great opponent, does not show seasonal progress. Ending the season on a high note is very important, especially for the best teams, which is why only the 1st & 2nd losses are counted.

So what’s so great about this rating system? A few things.

* It uses an easy to understand points scale: The more points a team has, the more successful they've been. Were I more math saavy, I might be inclined to try my hand at a more complex or technical system. But at the same time, there's something to be said for keeping it simple.

* It is tailored specifically for college football: This system can't be used for any other sport, or even for the NFL. It's designed for college football, making it more precise than if it was developed to rate multiple sports.

* It allows for exact comparison between teams and conferences of all years. Since the system is the same for all years, it is completely objective regarding time.

* Every teams starts out at 0 Points at the beginning of each season. There are no pre-season ratings to skew results, no built-in advantages for BCS conference teams, and no assumptions of power. All D1-A teams start on equal footing.

* The system doesn't spend any time looking forward, trying to be predictive. It solely measures how much teams have achieved thus far in the season, not how well they might do or how powerful they should be.

* Finally, if it had been the sole decider of who played in the BCS championship games from 1999-2007, it would have chosen 17 out of the 18 popular consensus teams.

The next page shows the Top 5 teams in each of the BCS years according to the my rankings.

Top > the BCS Years

Note: My system utilizes the information & scores kept by James Howell at his website. Thanks for all the hard work, James.

8 comments:

Anonymous said...

This does not take into account when teams take out their first string players in a "blowout" game (see Ohio State vs. YSU). There is NO way that Ohio State should be ranked #53 in the country...that is laughable on so many levels.

Sean said...

Anonymous:

These rankings are not biased based on the name recognition of a program as human polls are. Furthermore, they only take into account one week of games. As each team plays more games, it is much easier to judge the relative strength of each team.

Basically, if Kent State walloped Youngstown State at home the same way OSU did, I can guarantee you would not be laughing at Kent State being #53. Just give the rankings a few weeks to stabilize.

Ed Gunther said...

Thanks for the comment, Sean - couldn't've said it better myself.

Ed

riffraffDawg said...

When Vandy and Kentucky are the two highest ranked SEC teams there's a problem with the system.

Ed Gunther said...

As of week 2, 2008, Vandy and Kentucky are the only 2-0 SEC squads to have large MoV road win. What's so hard to understand about that?

I can't emphasize it enough - this system isn't about who's the best, or who's the most powerful. It's about who's achieved the most thus far in the season. In terms of what they've done, those two have achieved more.

Hans said...

Ed (or another familiar with this site), where can I find an explanation of what the abbreviated column headings (on your rankings tables) mean on the site? As in, what they stand for, and (if there is a calculation involved) how they are calculated?

Go Badgers!

Ed Gunther said...

Hey there Hans, there's now a link to a legend right above the rankings - hope that helps.

Hans said...

perfect, thanks!