computer rankings that make me wonder

[upperdeck]

Sagarin does some things that I just dont get and now that we are 2/3 of the way thru the season they should be coming together..

ND 42 2-5 sched 41
Mizzu 58 2-5 sched 34
MSU 68 2-5 sched 29
TT 48 3-4 sched 43
Miss St 56 2-5 sched 39
SU 72 4-4 sched 31

So I could say ND was better than SU on that day but over the course of the season with an easier schedule should it normalize closer?

if teams play what his system says are roughly the same schedule and has a better record shouldnt they be closer if not better in the ranking?

 

[HRE Otto IV]

The computer rankings are far from perfect but they are all we have. For whatever reason Bill Connelly's D rankings a well off IMO. BC has stunk against any O with a pulse this year. The rankings are supposed to take SOS into account. Yet he has them ranked as the 16th best D which is totally absurd. Which then overrates their overall ranking. Same thing happened last year with BC whose D was good but certainly not 3rd in the nation.

 

[orangeinjersey]

Rutgers at 114, 2nd to worst P5 Sagarin Ranked team (Kansas #1)

 

[GoSU96]

Sagarin does some things that I just dont get and now that we are 2/3 of the way thru the season they should be coming together..

ND 42 2-5 sched 41
Mizzu 58 2-5 sched 34
MSU 68 2-5 sched 29
TT 48 3-4 sched 43
Miss St 56 2-5 sched 39
SU 72 4-4 sched 31

So I could say ND was better than SU on that day but over the course of the season with an easier schedule should it normalize closer?

if teams play what his system says are roughly the same schedule and has a better record shouldnt they be closer if not better in the ranking?

Louisville putting up two games worth offense in one game has really skewed the defensive numbers. 845 messes with the average.

 

[Louie and Bouie]

Sagarin does some things that I just dont get and now that we are 2/3 of the way thru the season they should be coming together..

ND 42 2-5 sched 41
Mizzu 58 2-5 sched 34
MSU 68 2-5 sched 29
TT 48 3-4 sched 43
Miss St 56 2-5 sched 39
SU 72 4-4 sched 31

So I could say ND was better than SU on that day but over the course of the season with an easier schedule should it normalize closer?

if teams play what his system says are roughly the same schedule and has a better record shouldnt they be closer if not better in the ranking?

Believe it or not, most computer models don't take current record into account at all in their rating systems. Their models are a predictor of future performance based upon current statistical data to date while adjusting for relevant factors. In my opinion, the best computer rating systems take into account offensive and defensive efficiency (i.e. yards per play, points per possession) while adjusting for pace of play and strength of schedule.

Most of our computer ratings/rankings are currently in the 65-75 range which seems reasonable although I would rate us a little higher based on the "eye" test and the progression our team has made throughout the season.

I know its sacrilege to say on this board right now but it is the lack of points by the offense (despite all of the yardage) that is statistically holding back our computer rating right now and not our defensive production. Right now, we are averaging an adjusted 1.73 points per possession (26 points per game adjusted) on offense which puts us 81st place in the country. In order to be a Top 40 team I believe we need to average about 2.50 points per possession on offense (at our current rate of 15 possessions per game) which translates to about 37-38 points per game. Of course, as the defense gets better the stress on the offense to score 37 points per game will go down.

Just my two cents as I really like to look at the numbers behind the records.

 

[TheCusian]

Believe it or not, most computer models don't take current record into account at all in their rating systems. Their models are a predictor of future performance based upon current statistical data to date while adjusting for relevant factors. In my opinion, the best computer rating systems take into account offensive and defensive efficiency (i.e. yards per play, points per possession) while adjusting for pace of play and strength of schedule.

Most of our computer ratings/rankings are currently in the 65-75 range which seems reasonable although I would rate us a little higher based on the "eye" test and the progression our team has made throughout the season.

I know its sacrilege to say on this board right now but it is the lack of points by the offense (despite all of the yardage) that is statistically holding back our computer rating right now and not our defensive production. Right now, we are averaging an adjusted 1.73 points per possession (26 points per game adjusted) on offense which puts us 81st place in the country. In order to be a Top 40 team I believe we need to average about 2.50 points per possession on offense (at our current rate of 15 possessions per game) which translates to about 37-38 points per game. Of course, as the defense gets better the stress on the offense to score 37 points per game will go down.

Just my two cents as I really like to look at the numbers behind the records.

It's not sacrilegious at all. Most everyone here knows that there is a gap between yards/possessions and our points currently. You can see it in the stats and on the field - we've not seen the efficient version of this O yet.

The problem is when people think this is as good as it's going to get.

 

[HtownOrange]

The problem is when people think this is as good as it's going to get.

You win the the prize for the most obvious answer that few understand. HCDB is in week 8 of implementing a system that takes 18 weeks to be fully understood. We are on uphill climb, this roller coaster hasn't started rolling under its own power yet. That is when people will need to hang on for the ride!