Nice research. Very well done.
Busting The Myth of Option QB Fragility
One of the criticisms often heard of the spread option is that the QB in the system gets hurt with a higher frequency due to how many hits he takes. The idea does seem to make sense but does it hold up to reality? I’ve searched the MGOBLOG archives and haven’t seen a specific discussion of the topic so hopefully I’m presenting some new material.
The first thing to decide is an objective method with which to classify QBs without cherry picking players. To do this I generated a run-to-pass ratio by dividing total rushes by total passes. A high number means the QB runs a lot; a low number means he’s John Navarre. I then needed to decide on thresholds for running QBs vs. pocket QBs. After screwing around with the numbers for a while I came up with a Threat Level rating—not to be mistaken with Threet Level rating.
An R/P ratio of 0.5 or higher (1 run for every 2 passes) was a level 3 threat to run, 0.33 to 0.5 (1 run for every 3 passes) was level 2, 0.20 to 0.33 (1 run for every 5 passes) was level 1, and less than 0.2 is a level 0 threat to run. I admit these thresholds are somewhat arbitrary but they hold up to sense checks. Here are some notable QBs in each category from 2008:
Level 3—Dual Threat QB: Pat White – West Virginia, Tim Tebow – Florida, Terrelle Pryor – Ohio St.
Level 2—Running QB: Juice Williams – Illinois, Kellen Lewis – Indiana, Steven Threet – Michigan
Level 1—Mobile QB: Colt McCoy – Texas, Nick Sheridan – Michigan, Daryll Clark – Penn St.
Level 0—Pocket QB: Graham Harrell – Texas Tech, Sam Bradford – Oklahoma, Chase Daniel –Missouri, Mark Sanchez – So. Cal, Jimmy Clausen –Notre Dame, Brian Hoyer –Michigan St.
The next order of business was to filter out irrelevant players. To do this, I assumed that a QB needed to register at least 17 plays (any combination of passes and rushes) per game he played in to qualify as a non-scrub QB.
Then came the hard part: tallying the number of injured QBs in each category and the number of games lost due to injury then determining if there is indeed a difference between the four running threat levels. According to critics of the spread option, level 0 QBs should get injured the least often and level 3 QBs should get injured the most often. After a few solitary hours and a couple of six packs here’s what I found.
No. of QBs
QB Injury %
% Games Lost
Avg. No. of Games Lost
Surprise, surprise the critics are wrong. On a percentage basis the only group that suffered an out of norm injury percentage were level 2 QBs which I think of as QBs that are used like running backs (Juice Wiliams) or QBs that are too slow to be running in the first place (Steven Threet). All other groups suffered injuries at about a 23% clip. Meaning about 1 out of every 4 QBs in a given category lost playing time due to injury in 2008.
An interesting note that I did not expect to see is in the Average Number of Games Lost Column. Apparently the more stationary you are the more serious the injury you will sustain. This makes sense; in the pocket 10 linemen weighing about 300 lbs each surround the QB. In space, he can slide or go out of bounds and if he does get hit it’s by a 230-pound linebacker or a 200-pound defensive back.
Very informative: posts like this (in the style of Chris Brown at Smart Football) with lots of statistics and research are really interesting, much more so than most of the diary entries.
It's good to finally get that stigma of spread option QB being more likely to get injured debunked.
#3 category. If so, and the numbers held up, Sheridan could see more playing time than anyone but him wants. I like your work here, I hope you're right and Tate is able to avoid the big hit. If so, this could be a better season than anyone is predicting.
The point is not "Tate won't be more likely to be injured," the point is "a running QB is not more likely to be injured than a pocket QB."
There's a difference. For example, a running QB who is 6'5 and 235 is less likely to be injured than a running QB who is, say, 6'1 and 180. But the scheme has nothing to do with the likelihood of injury, IME.
I still feel nervous about Tate getting injured, but not because of RichRod's scheme, but more because he seems pretty fragile.
As far as Tate is concerned I am more interested in the stick figure quarterback injury rate.
I would like to see a follow-up where you take random year samples (every three years?) so as to debunk the "well that's just one year" argument. To me this analysis makes great sense. The guys who get killed are the stationary pocket passers who can't make the small lateral movements to get away (Level 0 guys). The Level 1 guys are the ones who can sit in the pocket and throw, or when the rush is present, scramble and get the five yards before going out of bounds/sliding/etc. The Level 2 guys analysis makes sense to me in that, as you point out, these guys are "fast" guys who may not be the best passers so that when they have to sit back in the pocket, the "pocket presence" isn't there and they get blindsided. Otherwise, Level 2 guys are slower players forced to play a running style (Chris Leak, Steve Threet, etc).
As an aside, if you wanted to do a study of extreme outliers, I would like to see an analysis of this type with Air Force/Navy QB's vs. Wisconsin/Jeff Tedford/David Cutcliff Qb's.
but I think when you stipulate that the sample should be "every three years" it ceases to be random......Just sayin'
Great work! Elegant and well written.
Also, it dispelled a feeling of disaster I was carrying around in my head that went something like this: I really think Tate is going to do better than people expect. He has a skill set that RR is going to know how to use. All those slants, rolling pockets, dump offs a la Noel Devine are going to mean a whole near gear for our offense. Of course, he will die and be crushed sometime between game two and game five because he's runnning around and will be killed. Please God, please spare him. But I know he won't be spared because he's going to get McGuffied sooner or later. How can he not?"
Such a public service is worth a big up arrow!
I think McGuffie got injured not only because of his well-documented "upright running style," but because he seemed to think he was an indestructible warrior of some sort, almost like he "marked out" over his own HS videos. He seemed to think that he could run and sustain contact with utter impugnity. The neurologists proved him wrong.
I can't imagine Tate Forcier being that stupid. And while I hope he is cocky like I believe a QB must be to lead effectively, I hope he isn't cocky in that regard. I'm sure he has already been coached to avoid or minimize the effects of contact, and will get even more coaching in that direction at UM.
We just need a sufficiently nasty escort to follow Tate around and ensure that he is treated with the proper respect.
Just tell him that he will lose in a head on collision with a linebacker at this level. Sliding or throwing it to tacopants are worthwhile endeavors.
I think McGuffie got injured because NO ONE blocked for him.
Did you see how many times he was nailed behind the line of scrimmage?
The upright style was a factor, but the poor kid had NO help.
By the time the line knew what they were doing he was already damaged goods.
He seemed very determined in the beginning of the season.
very frustrated during the middle of the season.
and very confused near the end.
If you know anything about Sam McGuffie, you should know that "cocky", and "stupid" aren't descriptive adjectives for him.
Way to be an ass.
Denard is almost certainly going to end up in the 'running' category, by how he is used. Based on the injury frequency of that category, we may end up using Gardner as a freshman, whether we want to or not.
I would also be interested in the results of adding a few years and a few teams, for more data.
Feagin? Is there a Threat Level 5?
And let's apply what you found with what we've seen. It holds up pretty good.
QBs at UM (after Bo) morphed into TE-sized brutes made to withstand the punishment. And fairly mobile recruits (like Henne) were bulked up to where they became immobile. To survive in the pocket, our system essentially required the QB to become immobile, robbing the team of a great weapon.
Those men took a pounding back there. They had to be huge. And still they could get hurt -- look at Henne's senior season.
Brady and Henson were the exceptions. Mallett was on his way to playing at 6'6" and 260.
Thanks for doing this analysis.
I'm wondering what might account for this seemingly counter-intuitive result. It seems reasonable to assume that 1) Most major injuries come through some sort of hard contact, and 2) QBs in your categories 3 and 2 are involved in more plays with hard contact. Given these assumptions, how does one account for the similarity of injury totals across categories?
My hypothesis is that QBs in your categories 3 and 2 are more "natural" runners, and more physically robust, than those in 1 and 0. Thus, if these same QBs were to just hang out in the pocket, they would be injured at a lower than average rate. Conversely, if you take QBs from cats 0 and 1 and put them in a scrambling/running offense, they would be injured at a higher than average rate. => It still might not bode well to have Tate carrying the ball 15 times a game.
I think you have something here. Remember Barry Sanders? Of course. Besides the fact that he frightened NFL defenders like headless zombies in the closet frighten 7 year olds, he demonstrated that runner's ability to deflect contact, to limit the impact of big hits. Natural runners approach contact and danger head up and thinking, thinking, thinking. They are the instigator and have a slight advantage.
Victims get destroyed.
I remember Warrick Dunn was being interviewed one time, when everybody in the NFL was gaga because he hadn't been ground into little bits after 4+ years in the NFL. Dunn pointed out he makes sure nobody ever gets a direct hit on him, and because of that he never really deals with injuries. In this way, Dunn said, his small size was his best asset, because it enabled him to avoid big hits.
Very interesting and a fun read. My theory on the level 2 injury percentage is that those are guys that run just enough to get themselves into trouble. As a qb, I think you need to either buy into running it full bore and just go for it, or take a safe approach and go down / slide / get out of bounds. When you get caught in the middle, that's when you get hurt. Kind of like playing not to get hurt seems to be the most assured way of getting hurt.
and am not totally suprised with the outcomes. In my mind (and experience playing football) moblie QB's actually seemed to get injured less due in large part to their ability to avoid tackles.
Also, when your feet are planted as a QB and you get hammered by a DL from behind your body takes a tremendous blow. If you think of it in terms of a car collision and imagine a car going at a speed of 60 mph hitting a vehicle at a stop sign there is far greater force exherted on that car than if it was moving in the same direction at 40 miles per hour when the car rearends it.
I think many of these #3 QB's will be getting dragged down from behind and hit from the side as they run which should lessen the impact of those hits IME.
I understand the physics of your car scenario, which makes sense, but I'm not sure the argument is that compelling overall, for a couple of reasons:
1) True, a mobile QB will probably have a decreased risk of being blindsided with his feet planted. Conversely, though, he'll have an increased risk of being hit head on while he and the defender are moving in opposite directions, which obviously increases the force of the impact (60 + 40 instead of 60 - 40, to use your example).
2) Being dragged down from behind might not result in any fantastic JACK'D UP OMG ESPN collisions, but it can just as easily tear a knee up like whoa.
The two points above are certainly part of the conventional wisdom that mobile QBs are more prone to injury (which the stats don't necessarily agree with), but I think they are legit thoughts/worries.
Much thanks to the OP for presenting the research, but I'm not sure you can conclusively say one way or another if a mobile QB is more injury-prone or not (yet).
The notion that pocket passers are somehow less vulnerable to injury is belied by the fact that every year, in the college ranks and in the NFL, pure, immobile pocket passers sustain serious, season-ending and career-shortening injuries. My bet is that if you took the time and tallied up all the pocket passers in D1A-level college ball who've gotten seriously injured in the last 10 years, the numbers would be substantial. Knees, shoulders, arms, ankles, skulls, brains, they're all at risk.
Rick Leach played for four full years. An option QB, he ran with the ball for a total of 487 carries over that time, and he played in every game during every season. If he missed any time due to injury, it was not serious enough to keep him out of a single game.
Steve Smith and Dennis Franklin would probably be in the Dual Threat category (not sure how effective Franklin was as a passer...), and both sustained game- and/or season-ending injuries running the ball.
Jim Harbaugh would probably fit in the Level 1 Mobile QB category, and he had his arm broken going for a loose football behind the line of scrimmage, IIRC. in 1979, although he had good wheels, John Wangler was a pure pocket passer. In the Gator Bowl, he was sacked (and cheap-shotted) in the backfield by Lawrence Taylor, and his knee was torn up on the play.
Football is a violent collision sport. There is simply no way to absolutely prevent injuries in general from occurring, nor is there a way to predict with any reliability whether a certain player or position is going to be injured or not.
still doesnt make me feel good about tate forcier possibility of going a whole yr uninjured
While injuries can happen in any play, the really worrisome hits on QBs are the blindside hits, esp. while standing in the pocket. These blindside shots occur most often with QBs who don't have a good sense of when pressure is coming from their backside. If you go back and review 1997, Brian Griese had an uncanny knack that year of sensing pressure when it was close to him, and he used his feet plenty to escape and buy more time. I think Henne had comparatively little ability to sense somebody closing in on him, and rarely used his feet to escape from trouble.
It was nice to read a diary that was well-written and had an idea supported with research. Thanks for taking the time to add something that I found valuable and well worth reading.
I remember suggesting that someone should look at the statistical likelihood of a running QB getting injured. I theorized that there would be no difference in injury rates, and a lot, and lots of people disagreed because OMGDennisDixon. BTW, you should email Brian with your results, because he is one of those people that concluded that "of course" running QBs get injured more often than pocket QBs.
So, naturally, I loved this diary (and yes, partly because I'm an "I told you so" dick). Good job, here's a point.
To solidify this conclusion even more, one could go back 3-5 years and see if the data still leads to the same conclusion. I imagine it will, but there will be naysayers who need more data to be convinced.
Your topic was well-chosen and the research makes good sense. Excellent post--thank you.
...but don't the arbitrary break points go a long way toward defining the analysis? If you combined the 0's and 1's you get an injury rate of 22.7% and combining the 2's and 3's you get an injury rate of 32% and your conclusions completely change (and combining the 2's and 3's to get comparable sample sizes probably makes sense here given the arbitrary nature of the cut points).
Considering the sample sizes of the 2's and 3's, I'm not sure you can really draw any conclusions from this. Moving two injuries from the 2 group to the 3 group would cause their injury percents to be very similar. What happens to the data if you make small changes to the cut points (instead of 1/2,1/3,1/5 maybe make them 1/3,1/4,1/5 or something).
The quasi-arbitrary break points worry me, too. Could we do a regression analysis to eliminate that issue? Any volunteers with plenty of time on their hands?
Actually, could we instead re-categorize Sheridan into the category most likely to not see the field? Any statistical fallacy you could arrange to make me think that he will never see the field would be greatly appreciated.
The data makes sense to me. What would be cool would be to compare this data to injury rates of running backs, because these mobile Qb's are a combo of Qb and running back so looking at both sides would be a better representation of the spectrum
The grammar police would have pointed out that "data" are plural.
(actually I'm just pointing this out as an example of paralipsis -- the bolded part -- for Meeesh...Dan)
Oh. Never mind.
Thank you for your analysis!
qSilver brought up a great point, though, in that the conclusions would change if Juice Williams et al. were in the same group as Pat White.
Also keep in mind that our pre-supposed running rate of Tate Forcier will likely be in that second group, since throwing accuracy, not elusiveness or footspeed is Forcier's 'weapon of choice.'
I don't want to knock you too much (no paralipsis) because I have a huge admiration for the work you've already put into this, but I'd like to see one more stat:
My thinking is that QBs who systemically run more often will, it follows, end up being on the business end of a tackle much more often. If a guy who gets tackled 35 times a game is less likely to miss time with injuries than a guy who gets tackled, say, 5 times a game, then we can rule out tackles as the primary cause of injury.
Therefore, we could surmise that the true enemy of QB health is not 'contact-made' (the basic premise behind the believe that mobile QBs are placed in greater danger) but 'type-of-contact.'
In other words, whether 'tis better to be brought down by a linebacker than sacked by a broken pocket, or leveled while standing in for a throw, or...(you know what's coming).
Our hypothesis should be that Pocket/systematically immobile QBs end up taking the 'wrong' kind of contact more often than scrambling/dual-threat QBs, or that the more time spent in the pocket, the greater the likelihood of injury. That will tell us, truly, if it's better to have linebackers, defensive backs, deathbackers, Schembacklers, or even chasing linemen hauling down your quarterback than leaving him squared to the chest pads of a group of beefy piledrivers (yes, it is coming).
Also, also, type of injury should be taken into account, and may help our case (which, ultimately, is to justify RR's system against claims that having your QB running around is a sure-fire way to lose QBs).
Because if there's any justification for loosing the opponents' defenders against our quarterback in space, as opposed to leaving him in the pocket and avoiding tackles at all costs, is that a player running toward expected contact is more likely to avoid an injury than a guy tasked with standing in until the following happens (it comes)
Hey. Let's not mock his lack of paralipsis. Having only one lip is a handicap, buddy, and making fun of the handicapped is not something we condone on this blog.
you thought about this and shit.
Rick Leach loves you.
now, something else to think about (perhaps if I'm bored one day I will do my own diary regarding this topic): the myth that spread offenses do not do as well in cold weather.
This is something I always hear about, but have no idea if it is just a false claim or has any substance to it. I would love to see a study on this. I'd do it if I had some guidance, anyone????
I don't think it's about the spread vs pro-style, but run vs pass. Yes, it's harder to pass a lot in cold weather because it requires more dexterity and precision, which both drop in poor weather conditions. However RichRod's spread offense is obviously based on the run. If anything, that should help Michigan in colder weather.
It seems to me that the idea of the spread offense automatically not working as well as the pro-offense (in bad weather) is nothing but Spartan and Irish mythology.
Thanks for doing that.