Tuesday, January 08, 2008

When Statistics Lie

For the second consecutive year Ohio State’s regular season statistics have been, in effect, a lie.

2006 Ohio State (regular season)

Scoring Defense – 10.4 ppg
Total Defense – 273 ypg
Rush Defense – 93.5 ypg
Pass Defense – 179.5 ypg

2006 BCS Title game against Florida (versus regular season)

Score – 41 (30.6 worse)
Yards Surrendered – 370 (97 worse)
Rushing Yards Surrendered – 156 (62.5 worse)
Passing Yards Surrendered – 214 (34.5 worse)


2007 Ohio State Regular Season Stats with national ranking

Scoring Defense – 1st (10.7 ppg)
Total Defense – 1st (225.25 ypg)
Rush Defense – 3rd (77.1 ypg)
Pass Defense 1st (148.17 ypg)

2007 BCS Title game against LSU (versus regular season)

Score – 38 (27.3 worse)
Yards Surrendered – 326 (100.75 worse)
Rushing Yards Surrendered – 152 (74.9 worse)
Passing Yards Surrendered – 174 (25.83 worse)

So in two consecutive “national title” games The Ohio State University has surrendered an average of 29 points more than their seasonal average. They have allowed 99 more yards than average, including 69 more yards rushing.

One is an aberration. Two is a trend.

As an avowed stat geek, this is of particular interest to me when I try to analyze games like this before they are played, as well as how it all figures into my Power Ratings which rely on season statistics for their accuracy.

My Power Rating for Ohio State (2nd overall) doesn’t appear to have been that accurate, as the stats on which it was based were flawed. Junk in, junk out as they say. And for the second consecutive year the Buckeyes looked better on paper then on the BCS Title field, as their inflated seasonal stats came home to roost.

There is one stat worth noting that, just perhaps, foreshadowed all of this – strength of schedule.

2006 Strength of Schedule

Florida – 1st in NCAA, 8th per Sagarin

Ohio State – 24th in NCAA, 38th per Sagarin (even after playing Florida)

2007 Strength of Schedule

LSU – 14th in NCAA, 11th per Sagarin

Ohio State – 39th in NCAA, 53rd per Sagarin (even after playing LSU)

For the second time Ohio State showed definitively it was not battle tested. The stats built on that Big Ten schedule were a lie.

6 comments:

jimcaserta said...

The problem is that strength of schedule should have a negative correlation with all other stats. For a given team, if you make their schedule harder, their stats should get worse. I feel, in handicapping individual games, too much goes into records. Teams don't change a whole lot with winning or losing, so a fluke play causing a loss doesn't tell you anything different than if that didn't happen and the team won. Playing tough games, and surviving/bouncing back, does build a confidence, that UF showed after going down 7-0 last year and LSU showed going down 10-0. LSU played a lot of the year from behind, and was comfortable there. In their losses, they pretty much played teir opponents even, while OSU got worked by Illinois. Tressel's magic seems to be gone too, with the bad time-out decision against Illinois, and questionable calls/discipline last night.

Anonymous said...

85% of all statistics are wrong...

Gator Duck said...

Actually, Ken, the statistics are not the problem. It is the interpretation of their meaning that is flawed. If only a reliable algorithm could be developed to combine all the individual elements, a reliable ranking system could be delveloped. That is the goal of Sagarin and others.

The concept of taking the results of various models developed by several sports crazed mathematicians and averaging those results is the closest thing we currently have, and it is obviously flawed.

I certainly don't have the answer. At this point nobody does. Mergz is making an effort to close the gap, and their are others as well.

Anonymous said...

Gator Duck - Agreed... I was merely pointing out the paradox of statistics. Much like how my wife tells me how much she "saved" at the mall :)

PS - I'm working up in Ohio area this week and it's amazing how they are already predicting that OSU will be in the top 5 pre-season next year...

Anonymous said...

IMO the SEC champ should keep getting voted into the title game every year until somebody beats them, and the Big Ten chump--I mean champ, should never be voted into the title game again. This should happen regardless of the two teams' records.

Hey, it's at least as logical as anything else the BCS has come up with.

Anonymous said...

For the second straight year, Ohio State skated through the regular season, barely encountering any adversity. They lost the only close game they played (Ill). Last season, no one really got close (the Mich final was misleading because UM tacked on a garbage score late).

There is no doubt in my mind that SEC teams, having been through the wars, are better equipped to handle the usual ebb and flow in a game in which both teams are very talented.

Everyone in the SEC except Miss. State, Ole Miss, and Vandy has plenty of speed. In the Big Ten, I'd venture to say that only OSU, Mich, and Penn St are stacked with high-level athletes. No other conference has top to bottom athletes like the SEC or is even close.