RB evaluation metric - UM's backs
I'm a big fan of the data posts on this blog, but usually can't do much myself to satisfy that craving.
This was some relatively simple thing that I'd noticed only being applied to the pros, and wondered about its validity for college players. I started collecting some numbers last year, but it sat around gathering dust. Brushing it off now in this content-challenged time before camp starts. Without further ado...
Is there any data to predict one back's success over that of another?
Bill Barnwell of the Web site FootballOutsiders.com formulated an equation that combines 40 time and body weight, and is an even better predictor of N.F.L. success than 40 time alone. Multiply a running back’s weight by 200, then divide that product by his 40-yard dash time, raised to the fourth power. The result will be a number that is, on average, about 100 for an N.F.L. running back, with big, fast players having higher numbers and small, slow players having lower numbers.
So basically (lb x 200) / (t ^ 4). Here is that math applied to Michigan's recent backs, showing how close they come to being an NFL back (who would probably be an awesome college back, controlling for OL yards, etc.)
CHART
De'Veon Smith (h) 4.50s 210# = 102.42
(C) 4.85s 223# = 84.01
(P) 4.59s 223# = 100.48
Ty Isaac (h) 4.45s 235# = 119.85
Karan Higdon (h) 4.50s 189# = 92.18
Chris Evans (h) 4.40s 195# = 104.05
Chris Evans (C) 4.64s 212# = 91.47
O'Maury Samuels (h) 4.48s 190# = 94.33
Kurt Taylor (h) 4.48s 200# = 99.29
Christn. Turner (h) 4.45* 187# = 95.37
Hassan Haskins (2) 4.74s 202# = 80.03
Eric Gray (h) 4.57s 191# = 87.57
Z. Charbonnet (2) 4.60s 215# = 96.04
Parentheses = source of data - (h)=Hello post (C) UM Combine (2) 247 (P) pro day (e) espn
Bonus math for the wallowers and what-if crowd:
Q. Crouch (e) 4.69s 240# = 99.28
Najee Harris (e) 4.66s 226# = 95.85
Kareem Walker (h) 4.50s 210# = 102.42
What am I supposed to get from this?
Well it shows that:
De'Veon Smith was likely to be the most successful of our recent backs to leave, which was borne out in reality.
Ty Isaac flamed out due to injury and his transfers. Can see why USC wanted him.
It seems to match star level as a recruit pretty well, as the guys over 100 were the ones that were 4 star or higher.
I think it shows that 40 times for high school recruits deserve more FAKES out of five.
Hassan Haskins was definitely under-the-radar for a reason.
A metric for pros is even more flawed than normal when applied to college kids.
I shouldn't write any more diaries unless I have better data and ability to format.
If there's any interest, I have thought of going back and trying to do classic UM backs since the Rivals era (1996-2014-ish)
It might help if you offered some anticipated scores for what is good/bad. For example, is Ty Issacs score of 119 a tremendous score, or is that just OK for a good pro? Are there good backs in the pros that buck the trend? Meaning, does this metric doom the careers of Higdon and nearly every recruit since? It would be helpful if there was more context for the numbers.
Saquon Barkley posted a score 124.33. His combine measurables were 233lbs with a 4.40 40 yard dash.
Sorry for the formatting, but I ran through some notables for comparisons. Emmitt Smith is the obvious outlier, being the all-time leading rusher.
Name, Weight, 40 time, Total
Chris Johnson, 203, 4.24,125.62
Karan Higdon 203, 4.40, 108.32
Barry Sanders 203, 4.37, 111.32
Saquon Barkley 234, 4.40, 124.86
Emmitt Smith 210, 4.7, 86.07
There are some problems here, notably the changing nature of weights through college and the fungibility of HS 40 times.
Higdon dropped a 4.4 last year in that combine, and he's a lot bigger than 189 now, for example.
I tend to think that this metric is more applicable to the NFL, where more things are equal, the players have gone through seasoning and training in college, and the measurements are much more specific.
Higdon was at 203 in June.
Adding to the methodology issues...all NFL players showed teams something that warranted signing an NFL contract. So not only are they big and fast, but they have other RB skills. When you look at Ty Isaac, he never consistently displayed skills other than being big and fast so his 119 score is not comparable to Barkley's 122
I'd be super interested in an updated rankings once fall roster is released, along with any updated 40 times known (such as Higdon at 4.4). I think our two top backs are likely much higher on that score than what you have.
Very interesting. I can think of two M backs right off the top of my head to both support and contradict this formula.
Tyrone Wheatley: huge with damn near world class speed. His M career lined up with what the formula predicts.
Mike Hart: small back with a (relatively) slow 40. Formula says not so good; obviously of M's best ever.
*one of M's best ever.
yes but this formula was made to predict NFL production, not college and clearly Hart did not have a great career, so the formula still holds for him.
I appreciate the effort that went into this post, but these numbers seem to be all over the map. Are we sure this is a useful metric?
Definitely not. Another one:
- Mike Hart: 206lb, 4.69s 40, ~85.15
I mean, it seems quite simple to say "if you're big and fast that's good". There's clearly a lot more than that which goes into being a great RB, as recent history at UM shows quite well (Mike Hart / Vincent Smith vs. Derrick Green / Ty Isaac).
Rashan Gary: (281lbs*200)/(4.57^4) = 129!!!!
Clearly Rashan should be our RB
Could imagine how badly Dantonio would shit himself if Harbaugh actually put Rashan in the backfield to take a hand-off against MSU in the game against them?
Probably burn his last 2 timeouts.
Interesting analysis, even if it requires some additional context/interpretation.
Plus I'm in favor of anything that will allow me to stop seeing the Herron de-commit diary.
A 140 pound cheetah running a 2.16 40 yard dash gets a score ten times higher than Saquon Barkley. Clearly we need to recruit a cheetah.
I would not want to tackle a cheetah!
Don't cheetahs go to the SEC?
Thank you for posting and I appreciate the effort to bring useful data back to diaries.
Unfortunately, this is not useful data. De'Veon Smith was not that fast and I would take Karan Higdon over him every day of the week--Higdon is faster and bigger than quoted here. Smith's true 40 is closest to the (C) time you have there.
Issac's time is way off. Evans is another productive back, and the second measure is probably accurate for him. Kurt Taylor is not a 4.48 guy at all.
Najee Harris has already demonstrated elite talent that far outstrips Kareem Walker.
I guess what this data demonstrates is that the metric is not useful for predicting college success, even if it has some relevance to the NFL.
FWIW, I believe Gray and Charbonnet are the most likely to succeed in college from that entire list--the only guy in their league is Higdon, and Evans right behind them.
Here's hoping Haskins or Turner can make an impact.
some more context on methodology: LINK to espn , LINK to 2018 Draft from Football Outsiders
Caveat from the author here: Garbage In, Garbage Out. This is only the data I could find from the sources listed, and if it's old or inaccurate, I can only blame what I copy/pasted. Obviously roster weights are snapshots in time (like boxing weigh-ins) if not entirely fictive.
Thanks for the good questions and the constructive criticism in your comments (and the cheetah joke, which I laughed at).
This seems like the critical passage from the Football Outsiders link:
"Speed Score measures speed in the context of strength and power. It doesn't measure agility, receiving ability or any of the other aspects related to the position. It does not claim that a larger player with a higher 40 time is somehow faster than a smaller player with a lower 40 time thanks to the power of exponentiation. Speed Score is useful because it's beneficial for a running back to be both fast and large."
It seems like the Speed Score could be a slightly helpful way of assessing whether a big RB has "enough" speed, or whether a small RB is fast enough to compensate for his lack of size--assuming, of course, that assumes the weights and 40-times you're using are reliable, that the bigger backs are heavier because of muscle rather than fat, etc. But at most this is going to be one data point in a sea of factors (vision, hands (both for ball security and receiving), blocking ability, attitude, football IQ, agility, stamina, and so on).
It would be one thing if having a high Speed Score was consistently correlated with being a successful RB and that having a low Speed Score was consistently correlated with a lack of success. But the examples given show both that some unsuccessful backs had high Speed Scores and that some highly successful backs had low scores. So it really tends to look like an arbitrary measurement IMO.
Nike tries to do this for HS kids with SPARQ scores and that includes additional data points.
Testing (SPARQ, combine, etc.) still is mostly useful as a tool to verify or add an asterisk to what you see on film.
The formula concept seemed pretty silly to me at first. I would expect a lot of safeties, linebackers and most of the larger wide receivers to grade out as amazing running backs.
I guess though, if you only compare running backs who already are NFL running backs and thus have the pre-requisite RB skills, the scores may have some correlation to success.
Still, Barry Sanders wasn't a big back and certainly wasn't fast for the NFL, but I've never seen any running back that was harder to tackle.
Yes. I think the formula would be better if it accounted for bulk, rather than weight. RB is one of the few positions where being tall can be a disadvantage. This formula favors them.
August 1st, 2018 at 12:39 AM ^
This diary is about as useful as the time I made a spreadsheet with every program's stance on satellite camps back when we thought there might be some sort of FBS-wide vote.
August 1st, 2018 at 12:57 AM ^
Having a fast 40 time is good but not a prerequisite to being a good RB. If you want a good athletic testing metric, you're better off looking at 20 yard short shuttle, 3 cone time, 10 yard split and broad jump. Anything less than 4 seconds in SS and 7 seconds in 3C are considered to be exceptional. As for 10 yard split, you want a RB to have that 10 yard split time of 1.5 sec or faster. For broad jump, anything over 9 and half feet are considered to be good.
All of the testings pretty much measure explosiveness, quickness and COD ability. Even then, RBs must have good vision or all that athleticism goes to waste.
Comments