DOD vs Snyder 2 (mini _FANalysis)

[kso_FAN]

There was some discussion in the other thread about the differences in the DOD era and Snyder 2, namely in terms of TOs and penalties, which I think is accurate. So the question becomes what else has changed?

I put together a chart that compares Plays per TO, Points per Play, and Yards per Play and another that shows penalties and yards per game. The chart only goes back to 2000, because I could only find accurate TO stats (which included fumbles) to that point. Still, it gives you a decent comparison between the end of the DOD and Snyder 2.





Snyder 2 has made a clear effort to reduce turnovers and penalties while the DOD had much more dominant defenses, though the last 3 years haven't been bad. The jump in yards per play allowed from now to then is dramatic and the jump in points per play allowed is significant. Snyder either a) knew the 2nd time around that he probably wasn't going to be able to collect the talent level necessary to be as dominant on defense or b) he knew that the nature of changes in offenses, especially in the new Big 12 and even after only a few years, wouldn't allow for the same type of dominant defenses. Probably, it was a combination of the 2.

Regardless, Snyder knew he would have to find some way to make up for those differences with his defenses the 2nd time around and perhaps that developed fully during the 09 and 10 seasons. As a result, he and his staff managed to put together offenses that maintained explosiveness and efficiency similar to his better DOD team while making improvements in TOs and Penalties. K-State turns it over much less on offense (and has in every season in Snyder 2) while having significantly fewer penalties, which offsets the defense allowing more yards and points per play than the dominant defenses of 00-04 (and 93-99).

[KanSt43]

I don't know what we do to deserve you....but well done.

[Brock Landers]

I had almost erased the 04 and 05 shitshows from my memory and then you go do this     

Just kidding man, great work as always.

[BIG APPLE CAT]

is "plays per game" available for this time period?  Would help normalize those numbers.

[kso_FAN]

is "plays per game" available for this time period?  Would help normalize those numbers.

[hemmy]

is "plays per game" available for this time period?  Would help normalize those numbers.

Everything except penalties is in a per play basis.

[BIG APPLE CAT]

is "plays per game" available for this time period?  Would help normalize those numbers.

Everything except penalties is in a per play basis.

the defense forcing a turnover every 27 lacks a bit of context without knowing how often they are out on the field.  One could also use plays per game to determine what impact, if any, pace of play factors into forcing/committing turnovers.

[CHONGS]

I think an interesting question to go along with this is: How much do turn overs really effect win %?

It seems like common sense that limiting turnovers is "good thing", and we often hear that getting more turnovers than losing them correlates strongly with winning.  Or least we are told this over and over again.

Interestingly enough I was looking at some teams that had a large number of possessions for each turnover they lost (which means they rarely turn the ball over).  I saw teams I somewhat expected like Oregon and Kansas State high on this list, but more often I saw teams like UTEP, Washington, Ball State, USC, Rice and Northern Illinois. A quick plot of all 128 team's win % as a function of possessions per turnover lost gives:



hmmm... not a particularly strong correlation at all (~0.43, where 0 would indicate no correlation and 1 a strong correlation [granted I am using a somewhat crude measure of correlation]).  Perhaps there is a sort of diminishing returns when it comes to limiting the number of turn overs you generate? At some point, the lack of risk taking on offense begins to limit your chances to win.

Ok then maybe it's more important to generate turn overs? I looked at teams that had the fewest possessions between gaining a turnover.  Here you'll find some expected named like Michigan State, TCU, Auburn, and Ohio State, but also a lot of Houstons, and Florida Internationals, Floridas and Virginias. A quick plot of all 128 team's win % as a function of possessions per turnover gained gives:



Here the correlation is even weaker (~-0.28 [the negative is because having fewer possessions between generated TOs leads to higher win %]).  Maybe this an effect of diminishing returns yet again, where having too aggressive of a defense (going for ints and strips) begins to limit your chances to win?

OK, then maybe it's really the net difference between possessions per TO lost and possessions per TO gained that will show us a good correlation with win %:



Not really,  the correlation here is strongest, but still only ~0.47.  Oddly enough, the teams that are still undefeated all hover around a net of 0.

Now what is a strongly correlated with win%?  That would be Net Points per Possession (Points Scored per Offensive Possession - Points Allowed per Defensive Possession):



This has a correlation coefficient of ~0.9.

[Skipper44]

I would say personel on this years offense is fairly similar to the 00 team, while I don't see any Newmans or Lebers on this year's defense the defensive numbers are surprisingly similar.

It is remarkable how 2.0 has reduced turnovers while increasing yard per play.

_FAN continues to be the BITBBS

[troubledscribe]

is "plays per game" available for this time period?  Would help normalize those numbers.

Has our time of possession gone up to decrease our offensive and defensive plays ?

[Skipper44]

is "plays per game" available for this time period?  Would help normalize those numbers.

Has our time of possession gone up to decrease our offensive and defensive plays ?

Snyder has always used the whole play clock, even when we were the finesse passing team in a run oriented Big 8.  The difference is now we play the majority of our games vs teams trying to maximize their offensive snaps and throwing the ball much more.

[puniraptor]

COMBO SEASON IS FINALLY OVER 

[kso_FAN]

COMBOTHE ROYALS SEASON IS FINALLY OVER 

with some 

[Fldermaus]

Here is another compelling factor to consider. In this analysis, I have plotted scoring differential, defined as the number of points scored by the opposing team subtracted from the number of points scored by the team on the X-axis, against the percentage of times that a team has won the game with that scoring margin. Fitting a 6th order polynomial against these data, we can see that there is a compelling relationship (r=0.62) between outscoring your opponent and winning a game. In the sweet spot around a 20 point differential, there is a nearly 150% chance of winning the game. The winning probability plummets between scoring margins of 40-50 points per game, which is why we see that winning teams try to avoid this trap zone. In large defeats, there is the moral victory spike, where fourth quarter points are scored against the opposing teams second string defenses.

Of course these data would be much more meaningful if normalized on a per possession basis, but son't be surprised to see that LHCBS will try to put the 'cats in a +17 to +24 point scoring margin over TCU.

[CHONGS]

[puniraptor]

Omg flder's photobucket

[yoman]

is "plays per game" available for this time period?  Would help normalize those numbers.

The fact that we have slowed our games down this much in an era where O-Coordinators want to run at least 80 plays a game, if not 90+ is impressive. Even moreso on defense, 65 plays a game is incredible. It makes sense why we are really healthy on that side of the ball for this late in the season.

[CrushNasty]

'02 should've been a natty ship   

[kso_FAN]

This. Good work Fldermaus.

[Cartierfor3]

COMBOTHE ROYALS SEASON IS FINALLY OVER 

with some 

could've done without the word "finally"

[Cire]

'02 should've been a natty ship   

One dimensional

[davegladow]

Comparing '10 to '11 is a weird experience.  I already knew '11 over-performed in terms of W-L (thanks mostly to turnovers, it appears), but to see it get its ass kicked repeatedly by '10 makes me question everything I thought I knew.

Just looking at yards+points, one could rank the teams as follows:

'04,'05,'09 < '10,'11 < '01,'13 < '00,'02,'03,'12,'14