Yards and points per play also are less relevant than per-possession stats. For instance Air Force will average a modest 5.5 per play which ranked 77th nationally. However their scheme is entirely based on getting consistent modest gains to put together long drives. Their yards per play would never indicate that they were 17th in yards per possession. They were a very good offense that simply hit few stat skewing plays. Air Force, Navy, and Army are all outside the top 40 in yards per play nationally. Army is 2nd in yards per drive, Navy is 11th, and AFA is 17th. Clearly yards per play got it wrong here as these teams are clearly moving the ball very well while intentionally shortening the game to bail out athletically outgunned defenses.Quote:
1- having a different opinion is fine but at least fact check and tell the truth. Worst defense in the history of Baylor football? 2011 & 2012 (the only years Bennett had this young of a back 7) were worse in both total D and scoring D. Simply untrue. They weren't good but why rely on something false?
Finally an intelligent mind to engage with! Total D and scoring D don't take into account pace of play so often it's not a good indicator when one teams is running 100 plays on defense compared to another only running 75. I like using opponents points per play because the number of plays at that point isn't as big factor (although still a factor). Baylor in 2017 was #110 in the country in opp pts per play giving up .501 per play. For comparison sakes, Baylor's often maligned 2012 defense (that gave up 70 points to West Virginia) gave up .455 per play.
Also teams that gamble on defense and take risks to get quick stops are disadvantaged in that statistic due to the disproportionate impact of big plays on the stat.
As for points per play? You not only have the quick vs patient style of play skewing but also the disadvantage of non-offensive scores or gifted points scored on "drives" that start deep in opposing territory. I find it to be a very poor statistic that needs a lot of work to get good value out of- possibly worse than the total points per game.
Quote:
2- Total Yards is borderline junk now due to different amounts of possessions per team. Yards per drive is a clearer picture. In terms of yards per drive vs power conference foes (because that's the most extensive set of per-drive stats I have) last year's defense trails only 2013, 2014, 2015, and 2005 in Yards per drive allowed out of the 13 seasons I have data for. When you compare them as a percentage of what other P5's get they only trail those teams and 2016 with 2005 and 2016 only being less than .6% of a difference.
I agree with you on the total yards thing because it doesn't take into account pace of play. I prefer opponent yards per play than yards per drive because (in my opinion) it takes into account pace of play much better overall.
Completely disagree on per play stats.
- Yards per drive DOES account for pace of play unlike total yards. I'd argue it does as good as yards per play with that. You won't accumulate bigger numbers by going fast with per drive stats or per play.
- It introduces problems of it's own with long plays skewing it as well as bias for and against styles of play.
Consider this example:
Offense has 4 three and outs that combine for 20 yards. On drive 5 they have an 80 yard touchdown. Total of 100 yards on 13 plays.
20 yards per drive would be a poor showing which is reflective of what happened but still shows they moved it some.
7.69 is a ridiculously high yards per play figure that if it was a season average would have been second nationally. Clearly they weren't dominating that defense.
Also it ignores STYLE of play as outlined in the service academy example. Once again, yards per drive tells the far more accurate story.
Quote:
The 2017 Baylor defense gave up 6.4 yards per play (#114 in the country). Again for comparisons sake, the 2012 Baylor defense gave up 6.0 yards per play. And they were dead last in total defense that season(again a junk stat, but used her to make a point). The only Baylor defense that compares is the 2011 one that gave up the same 6.4 per drive. Even Guy Morris defenses were better at this. On a side note, we were a much worse defense at home than on the road.
As my example above indicates, yards per play is easily dramatically skewed by long plays and to a lesser extent by style of play by a team and their opponents. I consider it to be a low reliability stat for that reason.
Quote:
3- 2016 allowed 38.7 YPD to power conference foes, 2017 allowed 35.8. Both allowed 2.19 yards more than other P5 teams averaged vs their opponents. This is after losing Blanchard, Levels, Stewart, Reid, and Edwards while losing X Jones and KJ Smith early in the year and constantly shuffling out injured guys. If this defense was all-time horrible- so was the prior year which I highly doubt you'd argue.
Again I'm not sure how much I buy the YPD argument. If you compare the Def S&P rankings used by football outsiders, the 2017 defense was significantly worse than 2016 (28.3 and #58 overall vs 33.2 and #111 overall). Again for comparison, the 2012 defense had a 31.5 rating and #86 overall.
S&P measures play by play success rates. It has value that I feel is more derived from their smaller breakdowns than their attempt to aggregate those into one metric. Some of these sub-metrics are better than others IMO.
For instance things like IsoPPP showing how often you make or give up explosive plays and havoc rate showing how disruptive a defense is, or stuff rate showing how effective you are at stopping runs for short yards play by play. Those are the kind of stats that carry a lot of insight to me and tells something descriptive that may not always be apparent with a typical box score.
I care less about their attempt to aggregate play by play efficiency. I like the idea behind success rate but it seems weaker to me. For instance Air Force in my example above probably doesn't trigger many "success rate" wins on first down but moves it ridiculously well per-possession. In their book a 4 yard gain on first down is success, this arbitrary metric doesn't see it that way.
Also they omit plays in the fourth quarter when the margin is more than 16 points. This clearly sometimes is a good thing but many times omits good data. For instance, Charlie Brewer vs WVU this year, Baylor's first comeback drive vs TCU in 14, etc. Games nowadays like UCLA/A&M that are far from over at 17 point margins with how many teams can go uptempo and get more chances with the ball. They have to draw the line somewhere, I just see any line there as too one size fits all.
I also lean to yards per drive as it's easily comparable, understandable, and is flexible enough to be used to answer multiple questions. It can:
-omit data that is irrelevant due to injury for key players (Very little real predictive value for WVU's upcoming offense with Grier can be gleaned from their bowl vs Utah without him)
-more easily isolate stats against particular styles of play, and it can show how scheme changes at half impact the game. (2016 OkSt vs Pitt is a good example as Pitts half two fire zones caused a dramatic YPD drop)
- It can be paired with field position stats to get a % if available Yards gained
- it can be paired with offensive scoring data to determine % of droves with a TD, FG, punt, turnover, etc.
It also allocates penalty yards among the offense and defense accurately when you take drive data from the drive charts.
I then use an opponent's games against other P5 teams as a fair "par" to measure against assuming they have a mostly P5 schedule. Holding Oklahoma to 30 yards per drive is great, giving that to KU is reason for concern.
Quote:
3- The defense did get better as the year wore on. I expect a better year with more experience and a better understanding of a more versatile scheme.
Agreed here. I just can't see how they could get worse. I actually have more hope for the defense improving than I do for the offense (although that isn't saying much).
Big key is the mental busts giving up explosive plays. That's huge and if they limit that they could take a big step forward. I expect far fewer as Snow didn't get where he is allowing simple busts to happen. The AAC was plenty wide open and they did a good job of bottling up explosive plays at Temple.