Started By
Message
Sev09's football computer rankings - 2013 season
Posted on 7/28/14 at 1:29 pm
Posted on 7/28/14 at 1:29 pm
Hey guys,
Being the computer/stats geek that I am, I decided to come up with my own objective rankings of college football teams.
Here are the results of my logic applied to the final stats of the 2013 season. I will be posting these computer rankings each week starting in week 6 of the 2014 season.
Let me know what you think - maybe I can tweak some things for accuracy. I'll answer any questions as well.
Being the computer/stats geek that I am, I decided to come up with my own objective rankings of college football teams.
Here are the results of my logic applied to the final stats of the 2013 season. I will be posting these computer rankings each week starting in week 6 of the 2014 season.
Let me know what you think - maybe I can tweak some things for accuracy. I'll answer any questions as well.
This post was edited on 7/28/14 at 2:46 pm
Posted on 7/28/14 at 1:30 pm to Sev09
Whats your formula?
This post was edited on 7/28/14 at 1:30 pm
Posted on 7/28/14 at 1:32 pm to Sev09
South Alabama ranked TWICE...
HOW AM I SUPPOSED TO TAKE THESE RANKINGS SERIOUSLY?!
Nah, other than that they look good.
What are your criteria?
HOW AM I SUPPOSED TO TAKE THESE RANKINGS SERIOUSLY?!
Nah, other than that they look good.
What are your criteria?
Posted on 7/28/14 at 1:32 pm to Sev09
I like how you came up with #2 at least
Posted on 7/28/14 at 1:33 pm to schexyoung
quote:
South Alabama ranked TWICE...
Toledo as well. Look good though.
quote:
Sev09
Requesting admission into the Brandon Harris fanclub
This post was edited on 7/28/14 at 1:34 pm
Posted on 7/28/14 at 1:34 pm to Sev09
Workout the formula or is hard to see where you went wrong, but clearly something is way off.
As bad as Arky was last year, they were much better than many of the teams ahead of them.
It appears your formula weighs win loss record too much without a corresponding weight given to strength of opponent.
As bad as Arky was last year, they were much better than many of the teams ahead of them.
It appears your formula weighs win loss record too much without a corresponding weight given to strength of opponent.
This post was edited on 7/28/14 at 1:35 pm
Posted on 7/28/14 at 1:35 pm to Sev09
it looks like SOS means very little and W/L means almost everything.
is USC ranked that high because they played the extra game against hawaii?
This poll is proof that W > SOS, so why should schools schedule anything over the bare minimum?
is USC ranked that high because they played the extra game against hawaii?
This poll is proof that W > SOS, so why should schools schedule anything over the bare minimum?
Posted on 7/28/14 at 1:36 pm to GEAUXmedic
Interested to know the formula too, especially considering how Bama is ranked #7 and Oklahoma #17, with the same record and OU winning head-to-head.
UCF had a better record than Baylor, and beat them.
I know it's a computer "power ranking," and can't take into account the actual games necessarily, but if you have a sec to determine what major criteria is being used.
By the way, I'm not disparaging your rankings, just bringing up talking points. Good work.
UCF had a better record than Baylor, and beat them.
I know it's a computer "power ranking," and can't take into account the actual games necessarily, but if you have a sec to determine what major criteria is being used.
By the way, I'm not disparaging your rankings, just bringing up talking points. Good work.
Posted on 7/28/14 at 1:36 pm to Sev09
It isn't terrible but obviously flawed. Baylor at 9 and UCF at 12 proves that.
Posted on 7/28/14 at 1:36 pm to autodd03
does it factor in miracles?
Posted on 7/28/14 at 1:38 pm to Sev09
Suspect...... just kidding.
How did you come up with your metrics? I think more people would take this a bit more seriously if they knew how you arrived at your claim.
How did you come up with your metrics? I think more people would take this a bit more seriously if they knew how you arrived at your claim.
Posted on 7/28/14 at 1:43 pm to idlewatcher
Sev I am a stats guy let's see your formula and see how we can tweak it
Posted on 7/28/14 at 1:44 pm to Sev09
quote:
What's your formula?
It's both simple and complicated at the same time. I'll start at the most granular level.
1.) Compute Rushing Power and Passing Power of each team. These are each calculated using important stats such as passing yards, completion %, Passing TDs, rushing yards, rushing tds, etc. MINUS interceptions and fumbles lost, times a multiplier, for passing and rushing power, respectively.
2.) Calculated average of Rushing and Passing Power to calculate Offense Power Index (OPI).
3.) Calculate Oppositions' rushing and passing power against, which measures the teams Defense Power Index.
4.) Defense Power Index (DPI) is then adjusted with additional considerations such as sacks, interceptions, forced fumbles, TFL, etc. for a small addition of "Defense Style Points". This correction takes into consideration the "eyeball test" and other physicality stats. After this correction, the DPI and OPI of the best teams are around 90-100 in either category, depending on their strengths.
5.) Power Index (PI) is then calculated as an average of the OPI and DPI. There are also certain measures in place, for example: If a team has a DPI of over 90, the DPI will be weighted more heavily in the PI calculation, since that is an outstanding number. Vice Versa for OPI > 90.
6.) After these detailed calculations, sort the teams by wins (descending), losses (ascending), and PI (descending).
7.) Then, I rank the conference's SOS by calculating the average ranking per conference, and assign relative Win/Loss multipliers.
8.) These Win/Loss multipliers are then applied to each teams' WL record.
9.) FINALLY, the final rankings are calculated as [Adjusted Wins] * (1 + (0.01*[PI]), descending.
These detailed calculations account for a team's true strength and capabilities. In theory, of course. I can't wait to see the results.
Posted on 7/28/14 at 1:45 pm to CockRocket
quote:
It isn't terrible but obviously flawed. Baylor at 9 and UCF at 12 proves that.
This is a strength ranking. The better team doesn't always win is all I can say.
Posted on 7/28/14 at 1:49 pm to Sev09
That might explain why the Big10 gets a lot of value in your rankings. Not wrong necessarily, but if the SOS of Big10 schools is "too high" then you'll get that fluctuation.
Posted on 7/28/14 at 1:50 pm to Sev09
good effort and it is much more sound than the "eyeball test" for sure.
Posted on 7/28/14 at 2:17 pm to schexyoung
quote:
South Alabama ranked TWICE...
HOW AM I SUPPOSED TO TAKE THESE RANKINGS SERIOUSLY?!
That was a hasty stitching together of multiple screenshots. Sorry about that.
Posted on 7/28/14 at 2:37 pm to Sev09
quote:
Sev09
Slow day at the office?
Posted on 7/28/14 at 2:46 pm to RGJ18
quote:
Slow day at the office?
This has been a couple weeks in the making - part of a bet between a buddy of mine.
Posted on 7/28/14 at 2:47 pm to GeauxPack81
quote:
quote:
South Alabama ranked TWICE...
Toledo as well. Look good though.
Fixed, btw.
Popular
Back to top
Follow SECRant for SEC Football News