Harbin Ratings for NCAA Division I FBS

As I mentioned in a post yesterday, if this doesn't prove I'm a football junkie and totally against the BCS system, I don't know what will, haha.

I started doing this last season and will likely continue doing this every year until we finally have a Playoff system in Div. I-A. For those of you in or from Ohio you should already be somewhat familiar with the Harbin Ratings System for ranking high school football teams. Well I adopted the Harbin System for NCAA Div I.

Under the Harbin system a points tier structure is established for the different tiers of play, in the NCAA this would be BCS conferences, mid-majors and then Div. I-AA opponents. A win over a BCS school is worth 5 points, a mid major 4 points, and a I-AA opponent is worth 3 points. These points accumulate throughout the season to comprise a team's Level 1 Points total. Level 1 Points are divided by the number of games played.

For example, let's say for the week 7 ratings (after 6 games played) Ohio State is undefeated. This means they would earn 4 points each for beating Akron and Toledo, and then 5 points each for beating Miami, Colorado, Michigan State and Nebraska. Ohio State's Level 1 point total would then be 28 points.

Level 2 Points are calculated by adding up all the Level 1 points of the opponents that a team has already played and applying a divisor to those points. For NCAA Div I-A the standard schedule is 12 games, so the divisor is 144 (12 opponents * 12 games scheduled). To adjust for an off or bye week, for each opponent that has had a bye or off week, 1 point is deducted from the Level 2 Divisor to adjust for whatever number of opponents have played one less game. For any I-AA team played it's also a 1 point deduction since the only play 11 games in the regular season.

Continuing using Ohio State week 7 for our example, say all our opponents we have played are undefeated, with the likely exception being Akron and of course Colorado after losing to Hawai'i, with the exception of losing to tOSU.

In this situation we would earn 0 L2 Points for Akron, 20 for Toledo, 18 points for Miami (one bye week), 18 for Colorado, 15 for Mich State (one bye week), and 21 for Nebraska, adding up to a total of 75 points and a divisor of 142 (2 point deduction for 2 total bye weeks).

The two modifications I added into this system are fairly simple.

First for each win over a top 25 rated team, a team earns 1 bonus point.

This factored in would give us 2 bonus points for our wins over Nebraska and Michigan State.

Second, a team earns a bonus for being rated in the top 25 determined by divided 1 by the team's rating. For example:
#1 = 1/1, #2 = 1/2, #3 = 1/3 etc.

For the week one ratings Top 25 bonus I use the USA Today / Coaches poll, namely because it's the only major poll where College coaches have a significant say in the rankings of teams. From week 2 and on I use the average of a team's Harbin rating and the USA today poll for that given week.

Say for Ohio State that at that point we are rated 7th in the Harbin system and ranked 5th in the USA today poll after beating Nebraska. Our current week's average top 25 ranking would then be 6th, earning us a bonus of 1/6 of a point.

To calculate a teams total points one uses the following formula:

(L1 points / games played) + (L2 points / L2 Divisor) + T25 Wins + T25 bonus

For tOSU in Week 7 in our example scenario it would look like this:

(28 / 6) + (75 / 142) + 2 + (1/6) = 7.3615 rounded at 4 decimal places.

Now this may not look like a lot of points, but I assure as you progress into the second half of the season the L2 points really start to pile up. Last year the Buckeyes finished 10th overall in the rating poll with 43.9306 points.

The beauty of this system is it takes into account the quality of conference a team plays in and both the quality of schedule and play by a teams opponents. It also factors in a a small manner the rating a team has each week.

Now bear in mind that this system is applied to a playoff system. Conference champions get in with a first round bye and then the remaining 32 playoff slots are filled with the top 2 rated teams remaining in each conference followed by the best 11 at large teams leftover.

It's not a perfect system, but I feel that it's an excellent theory for a compromise model in the meantime. It would allow 43 teams to be in contention for the National Championship, a true National Championship. And just think of the excitement and buzz that would come from the complementing of March Madness with a "Winter Wartime" or perhaps "Decisive December"?

After Miami and Maryland play tonight I'll compile the list and post the ratings for Week 2 and continue to do so throughout the remainder of the season. It should be fun to see what kind of Playoff Bracket would result at the end of the season!

If you're interested in a more thorough in depth explanation on the Harbin Ratings, here's the link to the OHSAA web page.

Football: Computer Ranking Examples

Originally Posted by McKraut

If you're interested in a more thorough in depth explanation on the Harbin Ratings, here's the link to the OHSAA web page.

Football: Computer Ranking Examples

There's a more thorough in depth explanation than that?

Originally Posted by PHATBUCK13

There's a more thorough in depth explanation than that?

I put it into about as close to laymen's terms as can be... they go into more specific details.

And here are the rankings heading into Week 2 of the season...


at the following link below. It was being a pain in the @$$ trying to copy and paste it into a post, so I'll just go this route.

https://skydrive.live.com/redir.aspx...04C4E268AC!123

I'm sure it takes about 4-5 weeks to work itself out.....

Syracuse #6
Northwestern #7
Houston #10
Ball St. #12
Hawaii #13
Nebraska #42
Mich. St. #45

One thing I do like is it takes the bias out and actually rewards teams for good wins.........We'll see what it looks like in week6........

I like it Sarge......

I am VERY interested in this! Could you post the Top 25 according to this system from the last couple of years? I'd be rather interested in seeing that. And do keep it coming! I'd also be interested to see if there was a way to adapt this to rank conferences.

Originally Posted by seismicmike

I am VERY interested in this! Could you post the Top 25 according to this system from the last couple of years? I'd be rather interested in seeing that. And do keep it coming! I'd also be interested to see if there was a way to adapt this to rank conferences.

I'll have to adjust the last several years to reflect the current system I'm using, I modified it slightly from years past, and I should be able to get the adjusted ranking from the last several years up sometime tomorrow. But the proof is in the pudding as they say, I can remember that my previous editions were a little more generous with the bonus points. But that in mind, to prove my point, in none of the previous years I did this did, Boise, TCU, or Utah end up in the top 10 at years end. Why, oh because they have fluffy conference schedules. Playoff teams? Absolutely, but not to the extent of them claiming to be ranked among the top 5 to 10 teams in the country.

@PhatBuck13 you are right, the initial 3 or 4 weeks are always much closer together until the strength of a team's schedule and opponents shows through. Week 7 or 8 is usually where you see the playoff picture begin to emerge, and around week 9 or 10 is once you've gotten past most teams' off weeks, so it really begins to balance out. But you are definitely in the right with the observation that it's designed to reward teams with quality wins and quality schedules, and of course, quality play.

https://skydrive.live.com/redir.aspx...04C4E268AC!123

Here's the updated Rankings for week 2. You can already see some very significant differences from last week...

Originally Posted by McKraut

https://skydrive.live.com/redir.aspx...04C4E268AC!123

Here's the updated Rankings for week 2. You can already see some very significant differences from last week...

Sarge-

When I clicked on the link, the week 1 rankings came up. Did I do something wrong?

I'm goin to bed, I'll check it out in the morning.

-phat