This diary was prompted by the debate from Tuesday between Brian and the Big Ten Geeks regarding the value of defensive rebounding. I read the Big Ten Geeks article that morning and had many of the same thoughts as Brian-I've never been a fan of the stops metric, particularly the way it was being used to compare players. As the debate moved to the value of defensive rebounding percentage, I decided to look through some Kenpom numbers to make a better argument for the importance/insignificance of that particular statistic.
|D-Eff Type||eFG%||TO%||DReb%||FT Rate|
A couple of notes. While I've labeled it as "DReb%", the statistic used was actually Opponent Offensive Rebounding %, hence the positive correlation with Defensive Efficiency (both statistics are "better" for the defense when the number is lower). TO% has a negative correlation because a higher TO% is "better" for the defense, so a high TO% would lead to a lower (read: better) defensive efficiency.
(It’s interesting to see how the Kenpom adjustments to efficiency change the numbers. eFG% and TO% consistently drop when adjusting for competition, while Dreb% and FTRate rise. The smaller deltas for this year makes me believe that this is a result of conference play and the leveling the playing field between teams that played non-conference schedules of varying difficulty, due to the relatively large proportion of non-conference game in the 2013 sample. I digress.)
It is well-known at this point that eFG% is by far the most important factor in defensive efficiency, but I was surprised that DReb% was the second most important factor (I had assumed it would be turnover rate). After seeing these results, I looked at the correlations between the four factors next.
So, there is a weak, but significant (with >340 samples) correlation between eFG% and DReb%. Going back to the correlations with defensive efficiency, I ran a partial correlation between DReb% and adjusted defensive efficiency, controlling for eFG%, which produced a value of…0.41. About the same correlation as TO% (a partial correlation for TO% is almost exactly the same as one without the adjustment, as you’d anticipate based on the low correlation between TO% and eFG%).
It looks like defensive rebounding is at least as important as the non-eFG% factors. What about the effect on the offensive end? Like Brian, I believe that steals should be valued more than other defensive statistics, so I went in assuming that we’d see some sort of correlation between TO% and Offensive Efficiency.
Negative correlations are due to lack of adjustment to defensive ratings for use with offensive efficiency (switching from lower = better to higher = better). However, from this, we can clearly see that defensive rebounding is just as important as any of the other defensive factors when it comes to offense. Michigan’s offense this season has shown this fanbase how defensive rebounding can trigger the break, but it is even more evident when you watch other leagues, where fewer teams put an emphasis on transition defense and sending players back on a change of possession and the game often breaks down into 2-on-2 or 3-on-2 runouts in each direction.
However, after all of this, I still believe that defensive rebounding is overrated as an individual metric. I'm not complaining about Jordan Morgan’s season, but he just isn’t a defensive game-changer in the way of Jeff Withey, Anthony Davis, Fab Melo, Nerlens Noel, or even A.J. Hammons. As has been stated, his high “stops” count is due to both Michigan’s excellent team Dreb% and Morgan's high individual number. His block and steal numbers are very low (his block% is 7th on the team, lower than all other starters, McGary and Horford). I might even argue that his presence has some effect on Michigan’s defensive philosophy and their inability to prevent three-pointers. With Morgan not a threat to alter shots inside, Michigan has to constantly switch on screens in order to prevent easy dribble penetration and 2-on-1 scenarios. They can’t fight over the top of screens to better challenge outside shots.
There was a great example of the effect that a shot-blocker has in the Iowa-Purdue game from Sunday, where Iowa’s players often had Purdue defenders trailing them after screens, but could not drive inside easily due to Hammon’s presence. Unfortunately, there weren't any Youtube highlights for that game, so I had to make due with the Michigan-Purdue game for an example.
First, Morgan sets a good screen for Burke. Hammons did not follow Morgan out to the perimeter, and you can see Ronnie Johnson start to fight through the screen at the top.
Burke is around the screen, but Johnson has followed him, preventing Burke from pulling up for an open three. Hammons is still in the paint, while Morgan is about to roll to the basket.
Finally, Burke has picked up the ball, unable to penetrate past Hammons or shoot over him. Purdue's defensive philosophy has helped remove the threat of a 3 from Michigan's balls-screen offense. Fortunately for Michigan, D.J. Byrd is still afraid of Burke and is about to jump in to help off Stauskas, leaving him open for a soon-to-be-bured 3. Not the best result for my example, but good for Michigan.
Further validating the importance of having a shot-altering presence: Correlation between block rate and defensive efficiency is very high (0.51), largely due to its influence on effective field goal % (correlation of 0.61).
This is all part of the bigger argument that the Big Ten Geeks make in their response to Brian's criticism-that post players/taller players should score higher on defensive metrics. Taller players can more easily influence defensive play away from their man, and playing on the interior puts you in better position for defensive statistics on every possession. Seeing as the objective of a perimeter defender is usually to prevent a single player from scoring/impacting the game, the best argument or evidence for an Oladipo or Craft would be to compare single game statistics vs season numbers for their primary defensive responsibility. They can’t impact the entire opposing offense and accumulate statistics in the same way as a Hammons or Berggren, but that’s a difference between the roles of perimeter defense and interior defense rather than a gap in defensive aptitude. You wouldn't want either of those guys I just mentioned chasing Trey Burke around the perimeter the way Christian Watford did, briefly, in last year's Indiana game. While Watford may have been successful initially, Burke got over the surprise and went on to score 18 points on 9 shots.
In my opinion, the best way to statistically evaluate individual defensive impact would be something similar to what Ace posted on Tuesday, evaluating lineups and considering an individual player’s ability to improve team defensive statistics while they are in the game. Now, this isn’t as fair to players like Caris LeVert and Spike Albrecht, who are rarely on the court with four other starters (theoretically the better defenders), but we could make an initial assumption that the other rotation players are all roughly equivalent when analyzing an individual player. It’s also unfair to players like Trey Burke, who might play 90% of the team’s minutes any given night and have a limited sample of largely garbage time minutes against which to compare the impact of their absence. That said, it would provide a better picture of a player’s ability to influence the opponent’s offensive strategy and results.
I am very curious to see the 3PA/FGA ratios and 3P% isolated for Michigan's three centers. Even though the team defensive philosophy remains the same for all three,it would be enlightening if opponents were taking more threes (or lower quality threes) depending on which player was protecting the paint. Ken Pomeroy wrote a blog post this week discussing the Syracuse zone and its (limited) ability to force lower quality three point attempts. Any effect at Michigan would likely be much smaller than that seen at the schools discussed in his post, but would still be worth examining.
Iowa normally averages 402 yards/game. Michigan held them to 302. For comparison, this is what Iowa had for each opponent:
|@ Iowa St.||365|
|@ 16 Penn St.||253|
Penn State is the only defense that held them to less. The defense played well yesterday. That is all.
I have been mulling over this for some time, so I figured I would ask the board (especially posters who are knowledgeable about recruiting) for some fresh perspective.
For the past couple of years, it's been a pretty consistent argument that defensive players will want to go to UM, in part, because of the availability of immediate playing time at key positions. The thinking is that a young player will see walk-ons and overmatched graduates on the two-deep and figure they can step onto the field as freshmen and compete for a starting spot from day one, as opposed to being buried behind a couple of experienced players. This makes sense intuitively, since it is obviously hard for a kid to go from being a star on his HS team to being a backup for 2-3 years, so the allure of immediately PT should be very enticing to stud defensive players.
Now, while I agree with this outlook to an extent, do people think the cratering of this defense the past few years might actually be working against defensive recruiting because good players don't want to be on a unit that is one of the worst statistically in college football, and has been a disappointment under RR? Beyond the obvious "good players like to play with other good players" argument, it seems even more important for defensive players to play on a good unit as opposed to offensive players - e.g. Calvin Johnson had a horrible QB, but he still gained recognition because one-on-one plays are "easier" for an offensive player than a defensive player. There are fewer one-on-one situations for defensive players (outside of DB in man coverage and maybe edge rushers), so a breakdown in the defense may lead to a players looking beaten on a play when in reality that is not true. And from a pure ego standpoint, why would a top-notch recruit want to come to a team that will give up 450+ yards a game and be subjected to criticism all the time, or want to play for a DC who might be gone in a year or two?
In this scenario, having early PT means you are also going to be thrown out there with a bunch of guys in similar situations (inexperienced and/or less-talented), and for the next 12-13 games you are going to be constantly highlighted as the weak link, chasing down guys on broken plays, etc. I'm not saying this is an insurmountable barrier for recruits - they might see this as an opportunity to turn the program around, to be the guys who brought back a great UM defense - but do people think it will have (or has had) an effect on the quality of defensive recruits?
In no way am I trying to question RR or his recruiting, but I am just wondering.
This is just going to be fairly brief, but I really feel optimistic about the outlook for this season. First, a few brief thoughts about last season. Michigan was about 6 inches away from 6-6 (Illinois), and a few "bad bounces" away from 8-4 (MSU and Iowa). With this view, we should all be thinking that 8-4 is certainly reachable.
Let's start at the very beginning. (A very good place to start!) Back at Center, we will see David Molk, who should be fully healed from his injury. Molk is potentially an all-conference center who can anchor the line. As far as the rest of the O-line goes, I'll let people smarter than me about these things tell you more, but from what I understand, it seems like signs are good that the team will be as good or better than last season on the offensive line.
I'm not going to weigh in on Denard vs. Tate at this time. (So what if that makes me chicken!) But at the very least, we should have a QB that is better than last season's QB. Tate is healthy, and with one more year of experience and work should be better than he was last year. Denard is beginning to look like a real QB, and not just a TB who occasionally throws the ball. By all accounts the sophomore version of Shoelace should be better than the freshman version of Forcier. So barring the double gut-punch injury to BOTH QBs, we should have a better QB than last year. Also, the competition this season between the two seems more active, which should make the current starter (whomever that may be) have to work harder to stay in that role. This means that the starting QB should be improving even more than last season. In the doomsday scenario, both of the best options get injured and we have (OMG) yet another true freshman playing QB in Devin Gardner, which--let's face it--is still better than a walk-on. In all liklihood, the starting QB at Michigan should be better than last season (and doesn't that make you feel better too?)\
Ok, so Minor and Brown (and Grady) graduated...and this is maybe a little sad. But Minor or Brown was hurt almost every week and Grady rarely saw the field. It seems like the important thing to have here is depth. And this Michigan has, perhaps even in abundance. There is some experience in this system for guys like Shaw and Smith (who should be healthy again). And this gives me hope that the status quo will be maintained and this group is at least as good as last season.
This year, Michigan has lost Matthews and Savoy (who I fondly remember for those catches against Notre Dame...happy thoughts!) But Hemmingway and Stonum are back on the outside. And Odoms and Roundtree will be awesome in the slot. Maybe Koger gives us a decent middle of the field threat, you know, the one we hoped he would be last season. There is plenty of reason to be optimistic with this skill group as well.
So to summarize, the offense should be better, in just about every phase of play this season. This might even make up for not having a space emperor (of space) winning the field position battle so often. We can always hope, right?
There can be hope with this group too. Briefly, can it really be worse than last season?
You don't lose Brandon Graham and get better. But Martin, VanBergen, Roh, Campbell, and Segasse make a good nucleus for DL depth. Who will step up? It is possible that the DL is still the strength of the defense this season.
Ok seriously. It can't get worse. Right? Please? If I close my eyes and wish really hard, does that make it so? Maybe this change in scheme we heard about is really about simplifying life for the LBs...and this can be a group that doesn't kill us on Defense. It's still the off-season, so I can be optimistic. Just ask any Lions fan you know (for whom the off-season, and sometimes, the preseason, are the best times to be a Lions fan).
Again, it can't get worse, can it? The big questions to answer. Will we have a reliable deep safety that doesn't get burned at least once per game? Will we have the athletes at corner to play adequate coverage? I think that there are enough hands on deck in that position group to be effective.
So in summary, the defense might even be adequate this season. Nobody will mistake them for one of the stellar defenses of the past (see 1997) but they should make do.
I think we can all crawl back from the ledge. I can't wait for September 4th to come!...And not just because it will be my 6th anniversary! ;)