Returning Production - ESPN
ESPN put an article out discussing how we can accurately (within reason) measure returning production. Here's what the author uses to measure returning production
It looks at the most predictive key personnel stats -- percentage of your QB's passing yards returning, percentage of your secondary's passes defensed returning, and everything in between -- and is weighted based on what correlates most strongly with year-to-year improvement and regression. It is a major factor in my annual SP+ projections, which will be released next week. (The other primary factors: recent recruiting and weighted five-year history.)
the article goes into future depth on things, BUT one thing I found interesting is that out of the 130 teams ranked, Michigan ranked 124th. On top of that, MSU (which regularly gets ridiculed around here for the rebuilding year they will have) is ranked above Michigan at 116. The teams below Michigan:
Kansas
LSU
Air Force
Alabama
UTEP
Utah
From a conference perspective, the B1G has 6 of the top 20 teams returning production. 6 others rank below 86th. Ohio State for perspective is 92nd, returning 56% of their productivity. Number 1 in the nation is Northwestern returning 84% of their productivity.
Some interesting snippets from the article.
Over the past six seasons, offenses with returning production above 60% average an improvement of about two points per game, while those below regress by about three. And the extremes are pretty stark: Only one of the 18 teams that have returned at least 90% of their offensive production saw its offensive SP+ rating fall, while nine improved by at least seven adjusted points per game.
Meanwhile, of the 37 offenses that returned 35% of their production or less, only five improved, while 19 regressed by at least seven adjusted points per game.
It's the same story on defense: Teams returning at least 85% of defensive production improve by an average of five adjusted points per game, while teams returning 40% or less regress by five adjusted points per game. If you're on one end of the spectrum or the other, your fate is pretty settled.
February 7th, 2020 at 1:13 PM ^
Can we get Shea another year ?
February 7th, 2020 at 2:30 PM ^
We have Chris Evan's back!
February 7th, 2020 at 3:55 PM ^
I liked Shea, probably more than most here, but I think that Dylan has that special something. I think our offense is going to improve this season, at least at the skill positions. O-line may not be as deep, but I don’t see a huge drop back to the (recent) bad old days.
February 7th, 2020 at 4:42 PM ^
I don't know that Dylan or any other QB on Michigan has "that special something" but I know that Shea didn't.
February 7th, 2020 at 1:14 PM ^
2016 was the chance.
February 7th, 2020 at 4:45 PM ^
Those are maybe once a decade chances. I hope it's not another 7 years before we get another realistic shit at a B1G title, but it wouldn't surprise me at all.
February 7th, 2020 at 8:33 PM ^
Looking back, that season hurts more and more.
Then again, at the time I just assumed Michigan was on an upward trajectory and would keep climbing.
sigh.
February 8th, 2020 at 12:02 PM ^
They were on an upward trajectory. Then Jedd Fisch left.
February 7th, 2020 at 1:21 PM ^
One things I think this is missing on (Among probably a bunch more other posters will identify) is last year we were breaking in a new offensive system. How much did that drag our numbers down last year and how much will it improve this year because of a consistent system. Also, I doubt that Bama and LSU fall off the map of being national contenders. It will be interesting to see if they do take a step back. Bama didn't win too many extremely close games where a loss of 2 points will make them lose this year.
February 7th, 2020 at 1:23 PM ^
From the article:
Percentage of defensive returning production formula derived from defensive line: 5%
Percentage derived from secondary: 37%
Percentage derived from full defense: 21%
I'm not understanding how this adds up to 100%. What am I missing?
6.7.0February 7th, 2020 at 1:33 PM ^
He didn't provide a complete breakdown of his formula/methodology, he just provided what he felt were the highlights. At least that is how I understood it.
February 7th, 2020 at 2:04 PM ^
He didn't provide a complete breakdown of his formula/methodology
Which is a problem.
The audience doesn't really have the context in which, say a Michigan (who is not a slouch talent-wise), is ranked at the bottom of some listing.
Yes, we ridicule Staee for their pending rebuild -- as mentioned in the OP. Rightfully so.
But since the article mentioned Shea and offensive line departures... and vis-a-vis Staee, nobody can tell me that either Dylan or Milton (even in their present state) aren't head and shoulders above Rocky Lombardi.
Rocky Lombardi fucking sucks. He has got to be the worst QB in a P5 conference.
February 8th, 2020 at 3:35 PM ^
Returning production isn't the whole of SP+ for a reason.
February 7th, 2020 at 1:24 PM ^
Does anyone think Northwestern's #1 returning "production" is really worth a damn ? I mean so they will be better than last year, but still that means they just stink less given the returners are not any good.
I think these kind of analytics mean virtually nothing in a vacuum. I mean Alabama is reloading with elite talent that others do not have in their rosters so being low in this stat probably means virtually nothing to their 2020 win/loss performance.
February 7th, 2020 at 1:31 PM ^
NW played nearly the vast majority of the year with a 3rd string QB. other than ohio (haskins), that's generally a problem for any team. see, e.g., o'korn, j. 2017.
February 7th, 2020 at 2:35 PM ^
All models are wrong, but some are useful --- as that old saying goes.
I think the author himself (Bill Connelly) would tell you it's far from perfect. But he does try to validate this method in terms of how it has done in previous years. And if he's telling the truth with his numbers, it does have some predictive value from a directional POV.
February 7th, 2020 at 2:44 PM ^
Leonardo DiCaprio finds models to be just right and useful.
February 7th, 2020 at 4:38 PM ^
...and young.
February 7th, 2020 at 4:38 PM ^
Oops... DP
February 8th, 2020 at 10:55 AM ^
Definitely. I think many here see a low number and assume that means he thinks we're going to win 3 games. He has 6 years of data that shows a pretty consistent outcome of production. It can't be 100% correct, but trending data shouldn't be ignored.
February 7th, 2020 at 1:24 PM ^
This is a metric I never really cared for? If you had a crappy offense (Northwestern), but return 84% of production... is that good?
February 7th, 2020 at 1:45 PM ^
They are using that as a basis to see if they improve or get worse. A bad team returning a lot of production will be slightly less bad (to the tune of ~2 points.)
February 7th, 2020 at 1:59 PM ^
It's a pretty spotty way of measuring what is coming back. Minnesota was super high last year and that was obviously a preview of good things to come because what they were returning was good. As you said, I really don't care if Northwestern is returning 84% of their production from last year, because they produced so little.
Sometimes this metric just doesn't pass the eye test. Losing your QB is always noteworthy, but whoever starts for us next year will have more time in Gattis' system than Shea did. We also return 87% of our rushing yards and 69% of our receiving yards. I get weighing the QB heavier than other positions, but I'm not worried about our returning offensive experience for 2020.
I think we return even more on defense, with the losses being Vert, Metellus, Hudson, and Uche. Uche, while a great player, was not getting a major percentage of snaps for us. Hudson was inconsistent - good when he was on, but made far too many mental mistakes in big moments. His greatest impact probably came on special teams at Michigan. Metellus, while very solid, is not an elite athlete. Vert is a tough player to replace. We lose Glasgow but get back Josh Ross. This is a defense that is really returning 7/8 of 11 starters, depending on how you look at it. Again, I can't look at this unit as being ranked in the 100s in experience.
I know production =/= experience, but we're dealing in the general category of "what does Team X have coming back next year" here. Michigan is a very experienced team in 2020 except for the most important position on the field.
February 7th, 2020 at 2:14 PM ^
I mean he doesn't say its impossible for a team with low returning production, its just extremely rare. So sure Michigan could buck the trend, but the reality is that Patterson was the best QB on the roster and the offensive line starters were above the guys replacing them. Given that its unlikely Michigans offense is significantly better based on returning production.
However Patterson seemed to regress and being in year two of the offensive system should help lead to a leap that overcomes those issues.
February 7th, 2020 at 2:24 PM ^
I understand what he's saying about using these numbers to project and the caveats that apply.
I'm looking at it from more of a binary analysis of "experience or inexperienced" and I would say that Michigan will be an experienced team next year. That idea does not comport with Bill's production numbers, which is why I think they're a bit of a flawed way of looking at experience. Sometimes the two align, but sometimes there's an obvious disconnect.
February 7th, 2020 at 3:41 PM ^
He isn't looking at experience, he is looking at production. Its better to bring back a sophmore who has seen the field and produced than slide in a redshirt senior to a position. Our team next year will be experienced in terms of years in the program, but they will have very little of proven production in an actual game, and we all know what practice hype is worth. So sure Michigan can buck the trend because we are plugging holes with guys who have been in the system so we are experienced, but historically replacing production leads to decreases in productivity regardless of how many years the replacement has been in the program. Talent plays, a redshirt senior usually isn't a super talented player, Alabama will sometimes start true freshman left tackles despite having nothing but blue chip recruits in front of them.
February 7th, 2020 at 4:18 PM ^
I understand what he is looking at, thanks. I'm not confused as to the difference between Bill's production numbers and team experience. I am saying that I don't think his production numbers are always useful.
It feels like you want to have a discussion about something I am not discussing. You seem to be under the impression that I am saying that Michigan's SP+ numbers will improve next year contrary to what Bill's production numbers project, but I have never said or implied that.
February 7th, 2020 at 4:04 PM ^
Did you use binary analysis when predicting Shea would be 1st team ALL BIG Ten last season ?
February 8th, 2020 at 1:57 AM ^
I mean the model is 100 percent helpful. Only one school bucked the trend on offense and 5 schools on defense. That’s a successful model. Look at a school like Texas. Preseason top 25, coming off a bowl win, plugging holes with a top 5 recruiting class and blue chips all over. Had a much worse year after being ranked 120th in returning production. It’s way more likely we look like Texas than have an amazing breakthrough replacing a starting qb, 3/5 members of the oline and our second best receiver.
February 8th, 2020 at 5:18 PM ^
Right. People just see Michigan low and say "this is a stupid stat", and ignore the part about how important a predictor it is. Turns out losing your QB and most of your OL is usually bad! There are good reasons to think it may not be so bad for Michigan though.
February 7th, 2020 at 6:02 PM ^
Replacing the entire interior of the Oline has to be taken into consideration for QB production. A new QB can be limited by potential issues along the Oline..And as of right now, the Oline is an unknown. It will probably be fine..but it can take time to gel and in the time it takes a new QB could be negatively impacted.
February 7th, 2020 at 3:40 PM ^
@michchip Why are you asking us if this is a metric you never cared for?
February 7th, 2020 at 1:26 PM ^
So it doesn't matter if your QB was Nick Sheridan or Joe Burrow as long as he returns you're good? I'm getting so sick of the advanced stat articles.
February 7th, 2020 at 1:57 PM ^
Joe Burrow was returning production. Now, you probably couldn’t have predicted the leap he took, but you could probably pencil in some improvement. It’s not that hard.
February 7th, 2020 at 2:02 PM ^
Every stat has to be taken in context, exponentially more so when you attempt to apply it to a specific player/team/situation. Generally, returning production correlates with improvement because the players returning have developed another year. As to your specific comment, it means statistically you are likely to be incrementally better the next year. Doesn't mean you'll be better than any of the teams on your schedule, just better than the team you were last year.
February 7th, 2020 at 1:27 PM ^
Anyone believe that crap?
February 7th, 2020 at 4:49 PM ^
Yes. These are not preseason ranking. It seems like they measure what they intend to measure accurately.
February 8th, 2020 at 5:10 PM ^
People seem to be taking this like he thinks Michigan will be in last place in the conference. It's just 1 measure. Michigan is likely to be around the top 10 because of the past 5 years and recruiting, and that sounds totally reasonable.
February 7th, 2020 at 1:36 PM ^
In other words, there is nothing of value to be gleaned from this little activity. The next survey will involve adding the numbers of the top players jerseys together and dividing it by the number of the worst player on the team...and the lower the quotient the better the team.
February 7th, 2020 at 1:37 PM ^
The more I look at our team I feel like there's two things that are going to happen. 1. This team is going to be really good and a lot better than anything we've seen under Jim harbaugh and really surprise a lot of people. 2. We're still going to lose to Ohio State and I record won't be much better than anything we've seen so far. I really hope that I'm wrong though on that second part ? and folks that's why we play the games... On paper were definitely losing that game in Columbus! But the game isn't played on paper!
February 7th, 2020 at 1:56 PM ^
I hate articles that don't include the quality of the returners. We are returning our #2 and #3 QB, our entire RB room save Tru, our #1, #2, #4, #5 receivers, our best receiving TE, and 1/5 OL. On defense we're returning our #1 and #3 DTs (3 might be low for Hinton, but i'm giving 2 to Dwumfour), our #2 and #3 DEs, our #1 and #2 ILB, our #2 and #3 CBs, and our #2 and #3 safeties. Looking at that it certainly doesn't seem as dire as ESPN is making it out to be. Losing the entire OL hurts, but if there's one thing the staff has done well on the recruiting trail it's stocking the OL room so we don't have any repeats of 2017.
February 7th, 2020 at 2:16 PM ^
I mean its all baselined to the team. He said high returning production leads to improvement. He didn't say Northwesterns offense would be good, just that its likely it would be better than last year
February 7th, 2020 at 2:18 PM ^
Returning production doesn’t take into account oline returning starters. It looks primarily at yards. So some teams may still have stellar numbers even with the loss of production. Unfortunately we are not one of those teams due to losing 4 oline starters. New qb new oline and a defense that will be average means another 8-9 win season. I’m sure the delusional fans are already preparing excuses for harbaugh next season. When will they learn...
February 7th, 2020 at 2:49 PM ^
I always chuckle at stats like this. They mean nothing. NW has the most returning production, but that production is low quality. It's just the opposite for teams like Alabama and osu, they lose a lot but they replace those losses with high quality players.
February 7th, 2020 at 4:10 PM ^
On top of that, MSU (which regularly gets ridiculed around here for the rebuilding year they will have) is ranked above Michigan at 116
The ridicule isn't just because they've lost a lot, but because they've recruited really, really poorly.
February 7th, 2020 at 4:17 PM ^
It's that quote that triggered me about this post. Look how stupid you Michigan fans are for not having bad players with game experience.