TSR's

Complete College Football Rankings

2005

---

Welcome to the second annual TSR College Football Ranking System by TheSportsRave.com.

The TSR system is designed to be as accurate as possible in the most comprehensive way using the average of six computers, the two opinion polls (coaches and writers) a point-scale system, which includes three columns of WIN points (WIN, OPPONENTS, OPPONENTS/OPPONENTS points) and three columns of SCORING points (MARGIN, OFFENSIVE and DEFENSIVE) and a column of POWER points.

This is mostly an arbitrary system and because there is no fixed ceiling for points to reach so there is no probability of distribution used. The TSR poll is designed so a team will gain ground off the strength of its victories. The deeper into the season, the more games played, the more points built up and the more accurate the system gets.

There are no predictors or complicated math theories and no points for losing to good teams. But the comprehensiveness along with the point-scale system where the amount of points scored, given up, and scoring margin is based on the ranking of the opponent, makes this system as accurate as any out there.

For the fun of it there will be two systems, THE MAIN POLL, where five computer polls (one was dropped after 2004) and the two opinion polls (coaches and writers) are calculated along with the point-scale system to start the preseason off.

The second, or ALTERNATE POLL, which is still being worked on, is an experimental one where last season's final computer and coaches/writer's points are averaged in with the following preseason computer and coaches/writer's poll points to start the season off. These two entities (computer and opinion polls) will be averaged all season long in the alternate poll as opposed to the normal way of "cold" starting. It will be interesting to see at the end of the season, which of the two versions is more accurate.

The five computer polls are: Congrove, Claasen, Matthews, Moore and Sagarin.

After the average is calculated, then a percentage is taken from that for this column's final points.

Along with the computer polls are the USAToday Coaches and AP Writer's polls. These two polls are added together then a percentage is taken from that sum for it's final points.

After one column of computer points and a second one of combined opinion polls, comes three columns: WIN, OPPONENTS and OPPONENTS/OPPONENTS points. This point-scale consists of all 119 teams divided into 11 groups (points given out for winning were raised for 2005 but the percentages taken from them were lowered for final poll):

1-5...........38/41
6-10.........35/38
11-15.......32/35
16-20.......29/32
21-30.......24/27
31-40.......21/24
41-50.......18/21
51-64.......13/16
65-79.......10/13
80-99.......5/8
100-119...2/5

Beating a team ranked (1-5) earns 38 points, 41 if won on the road. The next 30 teams are divided into three groups of 10: (21-30) through (41-50). The next 30 are two groups of 15 and then the final two groups of 19 (80-99) and 20 (100-119) teams.

All points are awarded in a cascading fashion--you get the most for beating the best. When a team wins, they get WIN points in which the amount depends on where the opponent is ranked at the time of the game (see scale above). The winning team then "owns" that opponent for the entire year. They then receive half their opponent's WIN points which is placed into their OPPONENTS category. And finally, the winner receives a fourth of the defeated team's OPPONENTS points, which goes into their OPPONENT'S OPPONENTS (Opp/Opp) point category.

A percentage is taken from each category where WIN points count the most among the three and the (0pp/0pp) count the least.

The next three categories are SCORING MARGIN, OFFENSIVE SCORING and DEFENSIVE SCORING. For these three, the scale is changed to groups of 10:

1-10......30/0
11-20.....27/3
21-30.....24/6
31-40.....21/9
41-50.....18/12
51-60.....15/15
61-70.....12/18
71-80.....9/21
81-90.....6/24
91-100....3/27
101-117...0/30
D1-AA.....0/40

The numbers on the right, (30/0), represent percentages. Playing a (1-10) team means 30 percent is added to what the team scored and for possible winning margin. The zero means no scale points added onto defensive points or a losing margin. Playing a team ranked (11-20) means 27 percent is added to offensive points and winning margin but now 3 percent is added to defensive points and/or a losing margin.

For example, giving up 20 points to a Top Ten team (1-10) means exactly 20 points are subtracted from a team's point total but against an (11-20) team, the 3 percent is added for a total of 20.6 points subtracted from a team's total points. If a team lost by 17 (to an 11-20 team), then 3 percent is added to the 17 for a total of 17.51 in the negative under SCORING MARGIN (which can be a plus or a minus depending on the quality of the team). The 17.51 would go into the SCORING MARGIN column and subtracted from a team's total points in addition to the 20.6 they gave up, which would fall into the DEFENSIVE SCORING column, which is also subtracted from the team's total points.

The scale for the last group, (101-119), is reversed, (0/30). This means no scale points added for beating a team ranked (101-119) but, 30 percent is added to defensive points and/or losing margin. Ten points given up now means a total of 13 points subtracted from a team's total when adding the 30 percent.

For D1-AA schools, the scale is (0/40). No WIN, OPPONENT'S or OPPONENT'S-OPPONENTS points are awarded for beating this opponent. The only points earned for beating a AA team, aside from computer and votes, is through offensive scoring and winning margin. But the penalty is steeper for giving up points, especially when losing.

Also, only 75 percent of a team's total in SCORING MARGIN and OFFENSIVE SCORING against I-AA opponents is counted as opposed to 100 percent against I-A teams.

The point-scales for the three WIN and three SCORING categories are arranged differently because it's easier to score on a team than it is to beat them. But the closer to the top five the tougher it gets, that's why I've arranged the WIN points-scale broken down to four groups of five for the top twenty.

Among the three SCORING categories, DEFENSIVE scoring counts the most against a team's point total. Next is SCORING MARGIN, which may be a plus or a minus, depending on the quality of the team. The least percentage added is OFFENSIVE SCORING, which is the least in weight because teams can run up the score after having the game in hand.

The ninth and last category is a POWER RATING. An average of all eight categories are taken before percentages are calculated, then a percentage is taken from that sum. The POWER RATING does not factor in until all teams have played at least two games.

An official blowout is 30 points and no more points are given to a team beyond that. Scale points are then awarded on top of the SCORING MARGIN, which also depends on where the opponent is ranked at the time of the game. A cap is also applied for offensive scoring, which is 50 points and scale points are added on top of that under the same guidlines as SCORING MARGIN.

Because of the format for overtime games and how easy it is to score, only the winning margin is calculated on top of the score after regulation. For example, a game that ends tied at 27 after regulation play with a final overtime score of 50-47, means the final score will be calculated as 30-27--the three-point margin on top of the regulation score.

So what you are looking at is an average of computer polls, the human element (coaches and writers polls), strength of schedule (WIN,OPP and OPP/OPP points), the three important scoring categories (SCORING MARGIN, OFFENSIVE and DEFENSIVE), and a POWER RATING. This system may be involved and dominated arbitrarily, but it's not complicated, which it doesn't have to be. The idea is to be comprehensive without being complicated, which the BCS hasn't figured out yet.

The beauty of this system is in the fact that it's able to start at the beginning of the season, which includes the fans. What makes up for any over or underrated teams early in the season, is taken care of as the season progresses: The more games played, the more points built up and the more accurate the rankings become.

A good example is if a team beats a 6-0 team that winds up being overrated by finishing 6-5. The winning team that beat the 6-0 team gets credit for that victory--WIN POINTS--based on the ranking of the 6-0 team at the time the game was played. But the winning team gets no more points off that 6-0 team (now 6-5) unless that opponent accumulates OPPONENTS points themselves. The winning team gets points worth those six victories as they "own" that team but that is all.

The winning team continues to feed off any opponent they beat through WIN, OPPONENTS and OPP/OPP's points, which may or may not be worth much at the end of the season.

The point being, who cares who's ranked high at the beginning of the season? That's for the fans to argue and enjoy but at least they're not excluded in the beginning, unlike the BCS system where the first rankings aren't listed until the middle of the season. So, the question is, how does the BCS become more accurate as the season goes along? They have a starting point as well, they just don't publish it.

Even though the TSR system is an arbitrary one, the numbers do come out as intended. For example the teams at the top of the rankings, I wanted the two opinion polls to count for a little more than the computer points and they do. The three WIN-point columns are to count for more than the computer and opinion polls for teams at the top of the rankings. And the three SCORING columns are to count for the least of the three sections of points.

With the exclusion of strength of schedule, the BCS has basically handed college football's fate into the hands of opinions. Sure, the computer polls count for one-third but the other two-thirds are opinions--coaches and writer's, which quite frankly, are the most arbitrary and least comprehensive of any system out there--and now, they are two-thirds of the BCS poll.

My theories and system can be compared by the full rankings listed below. The complete listings of the categories below, will be updated from time to time throughout the season.

Back To: THE TSR MAIN COLOR RANKINGS
Back To: THE ALTERNATE POLL

Filed by J.D. Long