Bang Your Head: Sanity in the Marshall, Dumervil, and Scheffler debate

Well I'm a stat-grinder, play-cruncher.
Now can we really keep our wide receiver?
Got no brains, he's insane.
McDaniels says that he's one big pain.
Brandon's like a laser, a 6-streamin' razor
He's got a mouth like an alligator.
But I want it louder.  Sack power
Should we keep Elvis when it strikes the hour?

There are a lot of opinions on Brandon Marshall, Elvis Dumervil, and Tony Scheffler these days.  Despite Marshall's verbal group hug during the Pro Bowl, the prevailing wisdom is still that the Broncos are going to look to trade (after tender offers) both Marshall and Scheffler, while making their best effort to retain the man-child sack leader in the NFL.

But is this the right strategy?   

Follow me after the jump and you'll find out once and for all.   For those that prefer the musical stylings of Quiet Riot, break out your spandex, hair spray, eyeliner, and click here.  

If you survived the jump, congrats.  You're about to enter a world that most fans simply never bother with.  You see, the simple answer to the Marshall, Dumervil, and Scheffler situations is to look at numbers of catches, numbers of sacks, and in the case of Marshall, the number of times he opens and closes his mouth, and go with your gut.  

And you know what?  There's no problem with it.  Often our first instincts are right.  So when you had a physical reaction to the drafting of Maurice Clarett or the signing of Dale Carter, you were spot on.  You didn't need stats or independent information.  You simply knew.

The current situation, however, presents a unique challenge.  Here we have three players.  The Broncos have to decide who of the three they want to keep, while trying to balance these contracts with value in other areas of need. All three players have produced (on the surface) at very high levels.   One of them, Scheffler, is presumed to be a bad fit for this offense.  The other is, for better or worse, the poster child for what some refer to as the modern athlete--selfish, petulant, and narrowly focused on individual accomplishments.  The last has generally been a good soldier--quiet, focused, and constantly learning.

So who stays and who goes?  To answer this question, we need to get to first understand the value of an individual player in the context of a common denominator with which we can compare all players.  We can do this through expected points values.

Points, Not Catches

In order to ascertain the value of each player, I recently charted every single offensive play from the 2009 Denver Broncos.  But even more head-bangingly delicious, I scored every single play with an expected points value.   That's right.  I can tell you what every play last year was worth, from the Stokley miracle touchdown (7.837 points) to Orton's pick-six interception against the Steelers (-8.067 points).  

Perhaps an example of how this works would be both fun and enlightening.  Ever wonder how many points Richard Seymour's hair pulling of Ryan Clady's bad-to-the-bone locks cost the Raiders?  Of course you did.  Let's walk through it.  Here is the simple description of the play from the NFL's official gamebook:

2-7-OAK 30 (12:04) (Shotgun) 8-K.Orton pass short middle to 10-J.Gaffney to OAK 28 for 2 yards (99-G.Ellis).

PENALTY on OAK-92-R.Seymour, Unnecessary Roughness, 14 yards, enforced at OAK 28.

Without the penalty, the Broncos would have been facing a 3rd-and-5 at the Oakland 28-yard line.  The expected points value of this down and distance would have been 2.882 points.  However, with the penalty, the Broncos were now facing a 1st-and-10 from the Oakland 14-yard line.  The expected points value of this down and distance was 4.922 points.  The differential between the two states?  2.04 points.  Thanks, Mr. Seymour!  You just cost your team the equivalent of a safety.   It's too bad you're going to be wasting away in Oakland Raiderville.

Now that we have a simple example, we can take our analysis even further.  We can use this kind of information from each and every play of the Broncos season to determine the value of individual players.

The Value of a Beast and a Scheff 

Let's begin with Marshall and Scheffler, simply because they are easy to analyze together since they are both pass catchers.   And, whether you like him or not, you have to admit that Brandon Marshall is a fascinating creature.  

After charting every pass play to Marshall and Scheffler (and every other Broncos WR), here is what I found:

Play Receiver Count Average Play Receiver Count Average
P Brandon Marshall 148 0.509 P Brandon Stokley 31 0.804
P Eddie Royal 77 0.044 P Knowshon Moreno 37 0.243
P Jabar Gaffney 85 0.535 P Correll Buckhalter 38 0.269
P Tony Scheffler 48 0.585 P Peyton Hillis 6 -0.111
P Daniel Graham 41 0.438 P Brandon Lloyd 17 0.326

Remember, these are the average points values of a pass thrown to each of these receivers.  So rather than simply including stats like numbers of catches or total yards (completely meaningless), these numbers convert an average points value to every pass thrown to these receivers.  So that route that Marshall didn't run hard against the Eagles?  It's in there.  The dropped screen pass from Daniel Graham against the Chiefs?  In there.   And so are all of their touchdowns.

Well, you get the drift.  We can convert all of these real game situations into an average points value per pass.    

So what can we gather from this chart?  Let's leave Eddie Royal and the others alone for a moment (I'll be dealing with them in another piece) and simply focus on Marshall and Scheffler.  Well, for starters, we can see that although Marshall was targeted a lot more often than every other receiver and he had some really interesting long scores, his average points per pass play were not through the roof (.509 points per pass).  Part of this is simply that Marshall had more passes with which to "smooth" his average.  But it's also hard to ignore the fact that Jabar Gaffney had a higher average points per play value (.535 points) than Marshall.  To this point we will return in a moment.

Scheffler, in slight contrast, had less targets, but a better value per play.  This should probably be expected given the types of routes that tight ends typically run in the NFL.  However, discounting Scheffler's accomplishments would be silly.  When the guy gets the ball on his hands, he creates value.  However, Daniel Graham, to my surprise also created decent value when he caught the ball.  So we don't necessarily need to get carried away when looking at Sheffler's numbers.  

What if we broke down Marshall and Sheffler even further by direction, by down, and finally, with a total points value:

Brandon Marshall                
Player Direction Count Average   Player Down Count Average
BM Short Left 51 0.45   BM 1 52 0.51
BM Short Middle 36 0.605   BM 2 56 0.299
BM Short Right 36 0.198   BM 3 38 0.717
BM Deep Left 5 1.19   BM 4 2 2.403
BM Deep Middle 9 0.629   Total Expected Points Count Value Category
BM Deep Right 11 1.075   BM 7 1.867 Running
BM Shotgun 91 0.456   BM 148 75.394 Receiving

Here is where things get interesting for Marshall.  In the short passing game, he did a ton of damage, averaging 0.60 points per pass.  He truly was a beast in the deep passing game, specifically along the sidelines.  And on 3rd down, he was extremely valuable to the Broncos, averaging 0.72 points per pass attempt.  This should not be discounted. On the most critical of downs, Marshall delivered. 

His total value to the Broncos during the 2009 season was essentially 77 points when we total all of his pass plays.

What about Scheffler?  Let's take a look at his chart:

Tony Scheffler                
Player Direction Count Average   Player Down Count Average
TS Short Left 16 0.322   TS 1 22 0.057
TS Short Middle 8 0.298   TS 2 16 0.973
TS Short Right 15 0.72   TS 3 8 0.697
TS Deep Left 6 0.673   TS 4 2 0.788
TS Deep Middle 2 1.714   Total Expected Points Count Points Category
TS Deep Right 1 2.273   TS 0 0 Running
TS Shotgun 22 0.603   TS 48 28.099 Receiving

Scheffler, unlike Marshall, tended to be more effective on routes to the Short Right of the field and he was simply deadly on 2nd down, averaging almost a full point per pass play.  Scheffler's total points value to the Broncos in 2009 was 28 points.  So although Scheffler might have had a higher value per pass play, the sheer volume of passes going Marshall's way gave Marshall almost three times more value.   This is another point to which we will return.  

Now we are beginning to get somewhere.  By translating the production of these players into points we have a common denominator with which to make a decision.  But we're not quite there yet.  Let's take a look at Elvis Dumervil.

The Value of Doom

Although I have not charted every defensive play for the 2009 Broncos (I'm in the middle of this now), I did take the time to chart the points value of every one of Dumervil's sacks and tackles, specifically for this piece.   Here's an example.  This is how Dumervil's Week 5 sack of Philip Rivers read in the NFL's gamebook:

3-3-50 (1:52) 17-P.Rivers sacked at 50 for 0 yards (92-E.Dumervil). FUMBLES (92-E.Dumervil), RECOVERED by DEN-99-V.Holliday at SD 47. 99-V.Holliday to SD 47 for no gain (68-K.Dielman).

The expected points value of a 3rd-and-3 from the 50-yard line is 1.39 points.  After the play was over, Denver had the ball at their own 47-yard line, facing a 1st-and-10.  The expected points value for such a down and distance is 2.019 points.  Since San Diego lost their expected points, we can add them to Denver's.  So the total value of Dumervil's strip sack was a whopping 3.481 points.   Or better than a Matt Prater field goal. 

After looking at both sacks and tackle throughout the entire season, here is how Dumervil's performance broke down:

Elvis Dumervil  
Play Average
Sack 1.456
Tackles, Passes Defensed, Other 0.231
   
Total Expected Points 33.076

The average value of all of Dumervil's 17 sacks, then, was about 1 1/2 points.   And when put together, the value of all of Dumervil's sacks alone were worth a total of 25 points.  His other tackles, passes defensed, fumble recoveries, and forced fumbles added another 8 points of value during the season.  

Admittedly, I charted Dumvervil's plays very quickly, so it's possible I missed a few.  Moreover, I gave Dumervil the full value of his forced fumbles on non-sacks, so the 8 points might be slightly high.  However, the major point remains.  Dumervil added the equivalent of about 33 points to the Broncos in 2009, or about 5 more points than Tight End Tony Scheffler.  Who knew that Dumervil's defense could be so offensive!  

I could go on to show all of Dumervil's point value by down, but everyone already knows Dumervil's propensity for sacking QBs on 3rd down.   However, when I finish my charting for the defense, I'll be able to do this for every defensive player and in each direction.  

Now we are at the point in which we can better compare the three players.  So let's line their total expected points values and points values per play side-by-side:

  Expected Points Value/Play Expected Points Value/Season
Marshall 0.509 75.394
Scheffler 0.585 28.099
Dumervil 0.612 33.076

How do you feel about the three players now?  Clearly we are blatantly ignoring the contribution from the other players on the offense that contributed  to the success of each of these players (the offensive line to Marshall's TDs, and Champ Bailey's coverage to Dumvervil's sacks), but we do have to draw the line somewhere.  

And already we can generate a few general conclusions, without the help of a straitjacket:

  1. Dumervil appeared to be slightly more valuable to the Broncos in 2009 than Scheffler
  2. Marshall, due to his sheer volume, had more total value than both Scheffler and Dumervil combined.
  3. Given Scheffler's higher expected points value per play, Denver should have probably thrown him the ball more often to generate more points.

Now I'm sure you've always wanted to pretend you are a General Manger of an NFL team, so put yourself in the shoes of Brian Xanders.  Each one of these 3 players is going to want to maximize their contracts, so the trick will be determining if the the contract they are seeking is clearly higher than their contribution to the overall expected points value of the entire team.  

In 2009, the total expected points value of the Broncos offense was 255 points.  We can't just assume that Brandon Marshall contributed a full 75 points (or 29% of the total).   As we've indicated, other players contributed to these expected points.  But once an estimate is made of the percentage contribution of each player to these points (and clearly Marshall and the QB would get a good share), a prudent GM can negotiate from a strong position.  

In this case (and admittedly a somewhat crude example), one might estimate that Kyle Orton deserves a quarter of these points, Marshall a quarter, the offensive line a quarter, and finally, the other backs and receivers a quarter.  This would put Marshall's true value at 18.75 expected points, or about 7% of the total expected points in a given year.   A prudent GM wouldn't want to--in any given year--inflate Marshall's contract to much higher than 7% of the total payroll.  GMs that do find themselves inflating individual player's contracts at premiums greater than the player's contribution to the overall team value quickly find themselves out of a job.  Or in the case of Matt Millen, they simply go to TV.

So, again, who goes and who stays?  By now, you know the answer is relative.  Xanders and McDaniels should tender offers that fall within a range of each player's contribution to the expected points value of the entire team.  If another team wants to beat these offers with some overpriced reach, the Broncos will be compensated, and then the question becomes, how to generate alternative investments that closely mirror the contributions of the departing players?  

Could you replace Scheffler and  his expected points?  Maybe not at the full 100%, but I'm guessing you could get close. Daniel Graham was responsible for 18 expected points himself this year, and he's not considered an exceptional receiving tight end.    

What about Marshall?  Probably not.  But a replacement at wide receiver (depending on the trade) could generate perhaps 60% to 75% of Marshall's contribution.  Jabar Gaffney is proof of this.  He contributed the equivalent almost 46 expected points in 2009.  And with Marshall gone, his points would only increase.  So losing Marshall, although painful, wouldn't be completely devastating. But it would sting for awhile.  

And Dumervil?  As one recent MHR member pointed out, sack artists don't grow on trees.  Replacing Dumervil's points would be a hefty chore indeed.  Robert Ayers won't even be able to get close to replacing Dumvervil's expected points for another year.  And the scary thing is that Dumervil is still learning the 3-4.   Sure, he needs to do a better job of stripping the QB, but I don't think he's peaked yet. 

So letting go of Scheffler is bearable.  Letting go of Marshall is not preferable if Denver can avoid it and avoid overpaying.  But letting go of Dumervil is simply enough to drive you crazy mad.  

If that happens, we'll all be banging our heads.

I’m glad we had this talk.  Now, vaya con Dios, Brah.

Agree, disagree, just like us on Facebook and follow us on Twitter so I can quit my day job.

Mark it Zero

Featured Stories

Offseason Coverage