DMN Blog: Gerald Sensabaugh vs. Roy Williams

[AdamJT13]

Arbitrary stats like 'success percentage' are complete BS. If a safety is asked to guard the deep middle of the field he is going to have a lesser success percentage than someone that only has shallow coverage responsibilities assuming an equal completion percentage.

For example if a safety is targeted because a corner got beat and he has deep responsibilities then if he can keep it under 10 yards I would say that was 'success.' According to this stat it is not. If you only have the short part of the field then you are going to consistently only be targeted in less than 15 yards which will boost the stat. According to this metric giving up 5 yards over and over again on second down is 'success.'

I realize they do this to fill books but at the end of the day when you start creating completely new subjective stats is when I begin to disregard.

I agree. Success percentage isn't the best measuring stick for defensive backs. Yards per attempt is a better measure.

Also, times targeted isn't always a good measure for safeties because many of them don't have man-to-man responsibilities. A good center-field type safety will get to more passes than one without as much range, and a safety who plays a lot of man-to-man will be the primary target much more than a guy who plays only deep zones as help for the cornerbacks. Looking back through STATS LLC's targets from 1995 through 2006, many of the best safeties had the highest number of targets -- because they played the most (a significant factor that's often ignored), had the most coverage responsibility and got to the most passes.

None of this particular post has anything to do with Roy Williams, either. It's just about the stats and how they're used and misused. And while I'm at it, I'll mention again how disappointingly inaccurate Football Outsiders' stats are for defensive backs, a fact I discovered after wasting money buying their play-by-play charting for the 2007 season.

 

[peplaw06]

Maybe a deaf, jumb and blind kid who can only play a mean game of pinball.

Speaking of.... did anyone see that story on Todd MacCullough the Sixers center from the early 2000s who is trying to be a professional pinball player?

 

[Clove]

One is a criminal, the other's play is criminal. So they're about the same.

 

[NeonNinja]

Traffic citations make you a criminal?

 

[Funxva]

One is a criminal, the other's play is criminal. So they're about the same.

I don't claim knowledge of what it takes to re-register a gun. I know that in places like Washington DC, there are a lot of hoops that need to be jumped through to get one registered.

Hell, some people are late re-registering their vehicles.

 

[Doomsay]

None of this particular post has anything to do with Roy Williams, either. It's just about the stats and how they're used and misused. And while I'm at it, I'll mention again how disappointingly inaccurate Football Outsiders' stats are for defensive backs, a fact I discovered after wasting money buying their play-by-play charting for the 2007 season.

How are they inaccurate?

 

[TellerMorrow34]

Interesting numbers. I don't know if it means it should all add up to more plays made on the field or not but I can certainly hope it will.

Either way I'll still miss Roy cause, well, yeah I'm a fan of Roy. So

 

[AdamJT13]

How are they inaccurate?

A variety of reasons. Too many passes didn't have a defensive back listed (even though you can watch the play and clearly see who it was), too many had the wrong player listed, the numbers in the book don't always match the numbers in their own play-by-play charting, etc. I don't expect 100 percent accuracy, but the combination of missing information and too much inaccuracy was very disappointing.

 

[khiladi]

If there is anything I'm not really worried about on this team, it is the defense, especially with the addition of Igor and this new guy, Sensabaugh, based upon what I see statistically. I think Igor is going to be better than Canty, and our coverage just got a lot more flexible.

 

[Doomsay]

A variety of reasons. Too many passes didn't have a defensive back listed (even though you can watch the play and clearly see who it was), too many had the wrong player listed, the numbers in the book don't always match the numbers in their own play-by-play charting, etc. I don't expect 100 percent accuracy, but the combination of missing information and too much inaccuracy was very disappointing.

Interesting, I compared 2 games to the official NFL game book + looked at every pass. Seemed like the 2 sources were in sync in terms of players involved yards gained etc. The broken coverage citations seemed a little arbitrary though. I haven't tried to compare their stats to an actual game replay though.