Wednesday, June 4, 2014

Hubba, hubba, hubba! Money, money, money!

If you were going to pay a quarterback $126 million over the next 6 years, with $61 million of it guaranteed, who would you pay it to?  Let's blindly look at the cumulative 2012-2013 statistical production of two very different quarterbacks.  Yes, this is basically like the Dating Game.




      GS*      Comp. %         Yards          YPA   TD Rate      Int Rate      Sack Rate
QB #1 16 62.54% 4590 7.88 5.67% 1.20% 8.24%
QB #2 23 59.77% 5011 7.90 4.88% 1.73% 8.67%

*GS = Games Started

I'm not saying that there is necessarily a correct answer to this question, or that I even care about these sorts of  things.  I just find it interesting how casually teams throw about such obscene amounts of money, with little regard for the long term consequences.  It's especially amusing when one of these two quarterbacks was a rather highly drafted  prospect, while the other was a bit unheralded and couldn't even get his coach to commit to him as the team's starter.  Can you tell which player is which?  Who would you prefer to have on your team?


Sunday, June 1, 2014

C.J. Mosley: Are Your Fingers Crossed?

Last year, I wrote a little post expressing my reasons for concern regarding the Ravens' selection of Arthur Brown, in the 2nd round of the 2013 NFL Draft.  It probably didn't make me terribly popular amongst the local Ravens fans.  I was shunned by my family, and forced to flee the state of Maryland.  Actually, I doubt that anybody noticed my little rant.  While I can't say with any absolute certainty what will become of Brown, I have to take the Ravens' recent selection of C.J. Mosley, who plays the same position, as a bit of a sign that the team may now be hedging their bets a bit. 

Now, at the time that Arthur Brown was selected, there was certainly a fair bit of hype surrounding him.  On Yahoo Sports, an article discussed why Brown could be a defensive rookie of the year candidate, before he even played a single snap, and comparisons were made between Brown and Luke Kuechly.  The Baltimore Sun discussed how Brown was a player that the Ravens coveted, and compared him to Ray Lewis, again, before he played a single snap.  Maybe Brown will end meeting these expectations eventually, I really can't say, though I have my doubts.

This year, similar hyperbolic claims are being made about C.J. Mosley.  On Ravens 24x7, the Luke Kuechly comparisons surfaced.  On the Ravens' own website, comparisons to Ray Lewis were somewhat cautiously thrown around, as we probably should have expected.  Yes, there have even been rumblings about how Mosley is already a strong candidate for defensive rookie of the year.  Again, this chatter has come before he has even played a single NFL game.  What could possibly go wrong?

While I may think this sort of speculation is a bit ridiculous, and feel a bit more pessimistic about Mosley's chances of success, I can see where people are coming from with this insanity.  If a middle linebacker is selected in the 1st round there is pretty much only one outcome that can justify such a selection, which is if the player becomes an absolutely dominant defensive force.  Teams don't select players at this position, this high in the draft, with the expectation of acquiring a player that is only going to be "good" or, god forbid, just "average".  They are expecting something a bit more outrageous, which is exactly the sort of expectations they should have for the price they are paying.

Of course, judging the "success" of a draft pick can be a bit tricky.  While I tend to lean rather heavily on a player's physical measurables, as well as their statistical production in college, to come up with a hunch as to what sort of prospect they really are, there are more powerful predictors as to what will happen with a player.

Based on CarAV (or, on a year to year basis, simply Approximate Value), the single greatest predictor of success for middle linebackers is probably their draft position.  Unfortunately, this is also my biggest gripe with CarAV as a statistic.  For the most part, it is a statistic that is just weighing a player's ability to get on the field.  Obviously, teams are going to start high draft picks, almost regardless of whether they are really any good.  As predictive statistics go, CarAV is just saying that the best predictor of success...is being predicted for success.  CarAV has some value, but for the most part is just handing out the kind of "participation medals" I got for playing on my 3rd grade soccer team.  There is a weird and disturbing sort of circular reasoning to CarAV that bothers me.  It doesn't really give us much of an understanding as to why these highly drafted players might actually be better.

Whatever the truth may be (you can probably guess what I think), it creates a worrisome question as to whether the best players actually get on the field, or whether players just wind up playing because teams believe in them for some unknown, and perhaps difficult to justify reason.  The pattern that teams have established of continuing to give repeated opportunities to former 1st round picks, who failed for other teams, rather than giving opportunities to less heralded players, is pretty obvious.  How else would players like Darrius Heyward-Bey keep getting opportunities?  When Kyle Boller failed as the Ravens QB, three more teams were willing to give him another shot.  Hell, the Browns even brought in Vince Young recently, for one more shot at redemption.  Is there a reason for this, or do teams just have a hard time shaking off their initial appraisal that such-and-such player was "destined to be a star, baby!"?

Regardless, I expect C.J. Mosley to get on the field, much as I said that I expected Arthur Brown to eventually get on the field last year.  If they get to play, I expect them to accumulate stats.  If they accumulate stats, I expect the fans to adore them.  That doesn't, however, mean that they will be great.

While it's fine to have your own subjective opinion on a player, this tends to be a bit of an unreliable way to make decisions.  I, subjectively, think I am the most charming man in the world.  Most women seem to disagree with me on this point.  Personally, I can't say that watching C.J. Mosley play made much of an impression on me, either positively or negatively, but there is really no reason for you to care about my opinion.  Your views may differ.  Instead of quibbling over "what we feel", let's instead look at what we can measure.

First, let's look at C.J. Mosley's athletic ability (based upon his combine results), in comparison to all of the middle linebackers who have been taken in the 1st round, during the last 10 years.  I will list the players 40-time, as well as their 10-yard splits, which should be pretty straight forward.  I will also list the player's Kangaroo Score (our measure of lower body power), which is given in the form of how many standard deviations a player is away from the average result for someone in their position group.  Since these players are measured along with heavier defensive ends and OLBs, their Kangaroo Scores tend to get pushed down a fair bit.  So, it should be noted that an average result for these prospects would probably be about  -0.800 (not just 1st rounders in this case, but amongst all player's at this position), while the average result for Pro Bowl and All Pro MLBs is closer to -0.400 (a small, but significant improvement).  I will also list the player's Agility Score (based on the 3-cone and short shuttle drills), which is also given in the form of how many standard deviations they are away from the average result, though no adjustments need to be made for weight class this time.


NAME         Year      40 Yard     10 yard       Kangaroo         Agility
C.J. Mosley 2014 4.63 1.56 -0.748 -0.251
Luke Kuechly 2012 4.58 1.56 0.303 1.256
Dont'a Hightower 2012 4.62 1.58 0.335 -1.397
Rolando McClain 2010 4.68 1.62 -0.564 0.074
Brian Cushing 2009 4.64 1.53 -0.219 1.133
Keith Rivers 2008 4.54 1.56 1.011 -0.092
Jerod Mayo 2008 4.54 1.50 0.142 0.034
Patrick Willis 2007 4.51 1.53 0.217 -0.282
Lawrence Timmons 2007 4.66 1.50 -0.481 0.702
Jon Beason 2007 4.72 1.60                 N/A             N/A
AJ Hawk 2006 4.59 1.56 0.421 1.891
Derrick Johnson 2005 4.52 1.61 0.070 1.223
Jonathan Vilma 2004 4.61                ? -0.376 1.516






AVG
4.60 1.55 0.009 0.483
MED
4.61 1.56 0.106 0.388


Who amongst these 13 players do you feel has been the most successful?  Feel free to make up your own mind.

When it comes to speed, everybody in this group produced fairly good results when it came to their 40-yard dash and 10-yard splits.  Only Jon Beason and Rolando McClain fell somewhat below the average results for 1st round middle linebackers.  I should probably note that Patrick Willis did run a rather stunning 4.37 second 40-yard dash at his Pro Day, though I am trying to limit this to combine results.  While Mosley's speed seems to line up with the average and median result for players in this group, there doesn't appear to be anything exceptional here for us to get excited about

When it comes to the Agility Score, you will probably notice that the majority of these players did quite well. Six, out of the thirteen players shown here, produced an Agility Score that was at least 0.500 standard deviations above average.  Five of them went well past even that mark, with results over 1.000 standard deviations above average.  None of them, with the exception of Dont'a Hightower, fell below Patrick Willis' result of -0.282, which is still basically an acceptably average outcome.  When it comes to agility, C.J. Mosley produced the 3rd worst result, though this is still close enough to average that I wouldn't write him off just yet.. 

Now we come to the Kangaroo Score, and its measure of lower body power.  While people might be surprised to see the celebrated Patrick Willis having such a mediocre Agility Score,  I would argue that he makes up for this with his Kangaroo Score.  Willis' result of 0.217 might not initially seem too impressive, but you have to remember that these players are being weighed against some larger defensive players up to 280#.  When we compare Willis' to the previously mentioned estimated average of -0.800, we find that he is actually about 1.017 standard deviations above average.  Even if compared him directly to a larger defensive end, Willis' lower body power would still be somewhat above average, and when compared to his lighter weight linebacker brethren, he is truly exceptional.  As I said before, the average Kangaroo Score for Pro Bowl and All Pro middle linebackers is approximately -0.400.  With the exception of Rolando McClain and C.J. Mosley, every player in this list seemed to meet or exceed this mark.  Mosley, obviously, easily did the worst of anybody in this group.

Amongst the players in this group, it is probably pretty easy to agree that Dont'a Hightower, Rolando McClain and Keith Rivers have probably been the most disappointing, relative to where they were selected.  That Hightower and McClain were among the least athletic in the group is something we probably shouldn't ignore.  As for the physically gifted Rivers, we'll get around to him in a moment.

So, despite the claims of Mosley's above average athleticism, there's really very little evidence to back this up.  On paper, he appears to be a rather average athlete in every category.  If your views regarding the relative quality of these players coincide with my own, I think we would agree that the most successful players on this list did generally demonstrate some significant physical advantages when it came to measured athletic ability.  Already, there seem to be some reasons to be a bit wary of C.J. Mosley's actual upside.

Of course, athleticism isn't everything, so let's look at Mosley's statistical production compared to his peers.  While all of these players have become middle linebackers (though Rivers might be an exception), some of them were used as outside linebackers in college, which would affect their statistical production to some degree.  Regardless, what we are going to do is list each of these players' statistics from their final year in college, but we will adjust all of their numbers to a 13 game per season standard, to level the playing field a bit.


Player           GP         TKL         TFL        Sack         PBU          INT            FF
C.J. Mosley 13 108 9 0 5 0 1
Luke Kuechly 13 206.9 13 0 3.3 3.3 0
Dont'a Hightower 13 85 11 4 3 1 1
Rolando McClain 13 97.5 13.4 3.7 3.7 1.9 0.9
Brian Cushing 13 73 10.5 3 6 1 1
Keith Rivers 13 84.5 5.4 0 4.3 0 1.1
Jerod Mayo 13 130 7.8 1.4 2.8 0.9 0.9
Patrick Willis 13 148.4 12.4 3.2 7.6 0 2.2
Lawrence Timmons 13 79 18 5 6 1 0
Jon Beason 13 99.6 10.3 2.7 1.1 0 0
AJ Hawk 13 131.1 18.3 10 2.4 1.2 2.4
Derrick Johnson 13 140.8 20.5 2.2 8.7 1.1 9.7








AVG
115.3 12.5 2.9 4.5 0.9 1.7
MED
103.8 11.7 2.9 4 1 1


While these results can appear to be all over the map, certain things do stand out.  For one, there is pretty much no statistical category in which Mosley produced anything that would fall significantly outside of the average range.  On the other hand, most of the more accomplished players in this list (at least highly accomplished during their college years), such as Kuechly, Mayo, Willis, Hawk and Johnson, did seem to show statistical production that rose significantly above the average in numerous areas.  Even if we leave out Kuechly's obscene tackle numbers, the remaining 4 players averaged 137.5 tackles in their final year which is 21.6% more than Mosley's result.  Some of the other highly accomplished players, like Lawrence Timmons and Brian Cushing, might have had a lower number of tackles than the average result, but brought a stronger resume when it came to high impact plays such as tackles for a loss, sacks, passes broken up, and interceptions.

I realize that many people don't like the idea of judging players based on their statistical production.  "The stats don't tell the whole story," you shout while gazing lovingly at your poster of Rocky Balboa, the eternal symbol of determination over quantifiable data.  Still, as Brian Burke of Advanced Football Analytics suggests when discussing his Tackle Factor statistic, there is something to be said for a player who frequently just shows up to make even a simple tackle.  Particularly with middle linebackers, who tend to be the greatest accumulators of raw statistical production due to their positioning on the field, it is something I wouldn't want to overlook.  While great statistical production may not ensure that a player will excel in the NFL, it is reassuring when a player has it on their resume.

You can also rather simply just crosscheck these two lists, and figure out whether a player was perhaps drafted due to significant athletic strengths (suggestive of potential upside) or due to their proven statistical production.  Players who did well in both areas, generally became rather high caliber players in the NFL.  Players who measured up poorly in both the athletic and statistical realms, generally appear to have become mediocrities.

Whatever your personal feelings are on C.J. Mosley, are entirely your own concern.  The question I want to ask is simple.  If Mosley is in fact just an average athlete, with a statistical history of performing at a rather average level, then what sense does it really make to expect him to become a dominant force when he moves to a higher level of competition?  If we discard our subjective feelings about Mosley, and just weigh him on the objective facts, how would we appraise him?  Personally, I would have a hard time betting on him turning out to be anything better than just an average player, and this is before we even discuss his injury history or the recent and dismal performances of over-hyped draft prospects from Alabama.

But...but...everyone says C.J. Mosley was the best middle linebacker in the 2014 draft, so I must be wrong about all of this, correct?  Maybe that's true.  I can certainly name some players that the computer had a low opinion of, yet managed to become quite good.  There are also a number of players that had very promising athletic and statistical profiles that failed to perform to my expectations.  Despite all of that, I have to go with what they odds say gives us the best chance of success.  To be perfectly honest, I'm not really sure that it matters whether C.J. Mosley becomes a success or not.  The real question that I think we should be asking is something very different.

Instead of criticizing the player, I really just want to criticize the thought process that went into selecting him.  If a team drafts enough players, with profiles that don't normally fit the conventional model of success, eventually some of them will turn out to be pretty good.  Eventually everyone gets lucky (though teams will suggest that it is skill rather than luck).  When the Ravens selected Arthur Brown in 2013, they clearly made this decision based on a purely subjective analysis, as there was little objective data to suggest that Brown was a terribly strong prospect (though the computer likes Brown just a tad bit more than Mosley).  Then, in 2014, they basically made the exact same decision about Mosley.  In many ways, the differences between the profiles of C.J. Mosley and Arthur Brown are virtually nonexistent.  By taking two such players, with relatively low probabilities of being highly successful, the team has probably slightly increased their chances of at least one of them being an outlier, who might be pretty good.  Despite that extremely small bit of deranged optimism, it seems to me that their odds of success would have been much more greatly enhanced by selecting someone who better fits the mold of historically successful players, and giving more weight to objective data rather than their scout's instincts.

Like I said already, I fully expect Mosley to get playing time.  Whether we, or the team, can accurately judge his performance once he is on the field, is going to be a bit debatable.  If the team analyzes their own players performance the same way they do draft prospects, in a rather subjective manner, it is doubtful that their ability to gauge these things is terribly reliable.  They will also be operating under the weight of confirmation bias, believing in Mosley's potential greatness, regardless of whatever may actually occur. 

If there's an upside to the gamble that the Ravens have taken on C.J. Mosley, it is that even if he fails to become "great", there will undoubtedly be teams interested in giving Mosley a second chance elsewhere.  After all, he was a 1st round pick, and all the evidence seems to support the idea that NFL teams never get tired of recycling players with a high draft status.  The Ravens even brought in Rolando McClain last year, after he disappointed the Raiders.  Second chances abound for these sorts of players, while later draft picks rarely ever get much of an opportunity.  Why bother giving some less heralded prospect a chance, when you can just place your faith in the general consensus as to who should become a star?  What could possibly go wrong with trusting someone else's opinion, rather than judging things for yourself?