Recent Trends in S&P+ OL Metrics
If you prefer to regard our beloved Meeechigan football teams through rose-colored glasses, then maybe you should stop reading here. Otherwise, if you have the intestinal fortitude to take a long, hard look in the cold light of day while the bombs fall around you, then by all means, keep calm and carry on.
Now that a couple of weeks have transpired since the Charlie Foxtrot that was the Orange Bowl, a consensus seems to have emerged that the root cause of many of Team 137's deficiencies is associated with the Offensive Line. Alas, this is not a new story, and the explanations as to how Michigan's OL has reached such a state are well documented. Nonetheless, some folks might still contend that the OL has improved under the tutelage of the new coaching staff, and any underperformance is simply a reflection of the available talent having maximized its potential. Maybe, but then again, maybe not.
Anyway, to get to the point of this post, Bill Connelly of S&P+ fancy stats fame now has available the 2016 season wrap-up metrics for OL performance. What's more, the same metrics are also available for the previous 2 seasons. So, after a bit of mousing around in the Excel, the following chart captures everything S&P you'd want to know about the OL's of Teams 135, 136 and 137:
To interpret the data, the pseudo-color scale is based on the ranking range of 1 to 128 (green = 1, yellow = 64, red = 128). Averages for the run-blocking and pass-blocking metrics are also calculated to give some sense of aggregate performance. The trend lines are keyed to the raw values, with green markers denoting the best performance values; red markers, the worst.
What these results show is a nearly wholesale regression in performance from 2015 to 2016, both in pass-blocking and run-blocking. All pass-blocking metrics have declined, with the PD Sack Rate being the worst over the past 3 seasons. As for run-blocking, SD Line Yards, Opportunity Rate and Power Success Rate are improved, but to start, SDLY and Opp. % couldn't have gotten much worse. On the other hand, Stuff Rate, PD Line Yards and Adjusted Line Yards all hit 3 year lows. In a nutshell, the OL shows qualified improvement in pass-blocking over Hoke's last season, but a general decline in the run-blocking metrics.
Just looking at the raw numbers is somewhat stunning in a few cases. A PD Sack Rate of 9.2% means that the QB is getting sacked about 1 in every 11 drop backs on passing downs. With that sort of sack incidence, it's no wonder the PD Line Yards are nearly the worst in the NCAA. Also, a Stuff Rate of 19.5% means that a TFL occurs on about 1 in every 5 rushing attempts. Poor damn Fitzgerald Toussaint has been replaced by poor damn Deveon Smith (and just nevermind poor damn Thomas Rawls).
To sum up, the bubble-gum-and-bailing-wire approach to installing an Offensive Line can only do so much and last so long. Talent, depth and experience make up the three-legged stool on which an Offensive Line Program is established. Alas, UM is still early in the process of building an OL Program. One or two five-star recruits do not an OL make, because like a chain, the OL is only as strong as its weakest link. A key injury can lead to disaster without depth to backfill and experience that can span the gaps.
Yours in football, and Go Blue!
January 13th, 2017 at 2:16 PM ^
I haven't formed an opinion on Drevno yet other than the obvious "In Harbaugh I trust", but this is dissapointing. I'm not saying he is bad, because he almost assuredly is not, but only losing one starter and regressing is not good. Neither is the fact that we lose 3 starters this year, possibly 4 ig Newsome doesn't come back. I still believe in Drevno as a good OL coach, but this is the first time I've had some doubts. Good job as always Ecky, really appreciate the posts.
January 13th, 2017 at 4:36 PM ^
Saying "losing one starter" is true, but, at the risk of telling just-so stories to justify by own biases, it's slightly more dramatic than that. By the end of this year, when the line was at its worst, we had new starters (for their positions) at LT (Braden), LG (Bredeson), and C (Cole). RT (Magnuson) was perhaps the most consistently accpetable position on the line. And who really knows what was up with Kalis at RG?
Anyway, my point is that Cole's move to center and (especially) Newsome's injury created a domino effect that I don't think we really recovered from. It's overstating the level of stability on the line to just say that it lost "one starter." There was more flux than that.
January 13th, 2017 at 9:39 PM ^
well then how do the stats bear out before versus after Newsome's injury?
As the season is now complete, the data would be even more reliable.
January 14th, 2017 at 2:00 PM ^
I thought that many of these guys repped at multiple positions so that if someone went down they can capably move to another position without a huge dropoff in their own performance, which may be easier than a first time starter moving somewhere new(er). Isn't that sort of the philosophy around "the best 5 start"? Shouldn't a 4th or 5th year guy be capable of moving from LG to LT or vice versa? I can understand a true freshman struggling anywhere.
But, I guess if the LG, in this case Braden, was only an average LG, then asked to move to LT, where he becomes below average, that makes sense. It's not like asking Steve Hutchinson to move from LG to LT. I guess it's just disaponting. Oh well. Hopefully Harbaugh can get the ship right in another recruiting cycle and year of experience.
January 13th, 2017 at 5:33 PM ^
January 14th, 2017 at 2:28 PM ^
Two things come to mind. One, there is more to moving the guy across from you than weight. Malik McDowell is 275 and routinely moves people back 20-50 lbs heavier than him. He is not particular big, muscular, nor has an intimidating frame for his position. Granted, all other things being equal, adding more body weight helps, but all other things are never equal and a guy that is stronger, more explosive, has better technique, is tough, and can identify who to block is much better than simply a larger guy that isn't as good at the other things. You won't find many better bodies than Braden at 6'6" and 335 lbs at a good body composition. He's also fast and strong for his position. But, he's still an average lineman. McDowell will eat Braden's lunch every time despite giving up 50 pounds of size and probably quite a bit of strength.
Second, dentifying who to block and getting to the second level is another problem we've had. Many of the o-line, and particularly Kalis it seems, really has trouble identifying who to block. Kalis is a big, strong, intense, and powerful young man, but he's not a great lineman because he never knows who to block, which slows him down and takes away his aggression. None of our lineman do a particularly good job at getting to the second level which is one of the reasons why we rarely break big runs.
January 13th, 2017 at 9:30 PM ^
This is a great post and I am not trying to attack it but... in 2014 our starting QB was Devin Gardner, 2015 Jake Rudock and 2016 Wilton Speight. Anyone who watched those games would know that a very high percentage of those sack rates listed could be explained by QB pocket presence (should come as absolutely no surprise that the sack rate went down tremendously after 2014 because Garner had zero pocket presence). If this data set were from an NFL team with a consistent starting QB and mostly similar O-Linemen then I would say you could attribute a rather high percentage of the variation to coaching. To attribute any of the change in these numbers to coaching is very shaky at best. This is just one example of how one statistic could be heavily biased. With so much turnover in personnel at the collegiate level I would say that the same argument could be made for all the statistics listed. OL data is so player specific that attributing changes to coaches is not very fair in my opinion.
January 14th, 2017 at 8:55 AM ^
January 13th, 2017 at 10:51 PM ^
against what we are able to do to opponent OL?
January 14th, 2017 at 2:20 PM ^
January 15th, 2017 at 2:30 PM ^
The 2015 offense averaged 4.2 ypc and 158 ypg rushing.
The 2016 offense averaged 4.8 ypc and 213 ypg rushing.
And yet, you're telling me that Bill Connelly says the 2015 line was better. That Tyrone Wheatley, Sr. must be a genius. Get that man a raise ASAP.
Something, something, smell test.
January 15th, 2017 at 4:38 PM ^
Many of Connoley's stats are adjusted for opponents.
Although it is a myth that Michigan merely beat up on inferior opponents and wilted against stronger ones as a team, this criticism held true with respect to offensive rushing performance. Said differently, the defense thundercrushed Rutgers, but also played great against Ohio State. But the rushing offense only looked good when the opponents were bad.
January 17th, 2017 at 12:05 PM ^
An effective OL will minimize tackles for loss, and get sufficient push and create the gaps required for any given ball carrier to get the yardage necessary to keep the chains moving. Any yardage beyond that depends more on the talent of the ball carrier to speed through gaps or around the edge, and avoid being tackled. Yet, YPC is a simplistic statistic that does not in any way separate the contributions of the offensive line, which are more critical to obtain required yards, from the ball carrier, which are more evident when reaching and exploiting the second level.
The purpose of the advanced metrics, such as Line Yards, is to give a better indication of the efficiency of an OL by weighting longer gains (for which the ball carrier is more attributable) less than shorter gains (for which the OL creates the space for the yards accumulated), with negative gains getting the heaviest weight of all.
Comments