Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2017 Offensive Personnel Analysis

It's a three-receiver league, but for the first time since 2010, the frequency of 11 personnel actually went down last year. Was it a blip, or sign of things to come?

16 Mar 2016

2016 Free Agency Cost-Benefit Analysis

by Andrew Healy

When the Jaguars signed free-agent defensive end Malik Jackson, they likely had visions of Muhammad Wilkerson dancing in their heads. In his best moments of 2015, Jackson played near the level of the Jets' 3-4 end, a two-time second-team All-Pro who is just three months older than Jackson. It's hard to think Jackson will be better than Wilkerson has been over the last three years, but squint hard enough and you can see a future where he just hits that mark. Maybe.

And that's the problem with the Jaguars' decision to make Jackson one of the richest men in football. His contract (five years, $85.5 million, $42 million guaranteed) will be a wash -- meaning, he will be worth about what they paid -- if he plays like the beast who appeared in glimpses in 2015 after three years mostly not even starting for the Broncos. The downside to the Jackson contract? Almost limitless.

If we apply cost-benefit analysis, the Jackson contract beats last year's Byron Maxwell deal to be the worst free-agency decision of the last two seasons. For the Jaguars, Jackson joins a parade of poor value propositions from last season. No team came close to the Jaguars last year in expected value lost by overpaying for free agents.

To analyze this year's contracts, I'll again be looking at the expected value of each deal on the day it's signed. I don't claim to be able to predict future performance with stats any better than talent evaluators can with scouting information. I'm just using a rough estimate of likely future value to evaluate the contract. Rather than the projection of future performance, the more interesting number in the contract evaluation is the benchmark the player would have to hit to live up to the deal. Consider the result for Malik Jackson:

Malik Jackson's Contract With Jacksonville
Player Pos Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Malik Jackson DE 26 JAC 5 $17,100,000 $42,000,000 7.78 13.68 -5.90

Based on his recent performance (according to Pro-Football-Reference's Approximate Value) and his age, I'm projecting Jackson to be a 7.78-AV player going forward. (For an idea of what that means, here is a list of players with exactly 8 AV in 2015. Jackson was worth 9 AV last season.) Jacksonville would certainly disagree with that projection, and they may well be right. But if there's a bunch of uncertainty about how well Jackson will play going forward, there is much less about how well he needs to perform to live up the deal. That key number above is 13.68 AV -- about Muhammad Wilkerson's average mark over the last three seasons. We can use these ideas to evaluate any free agent contract.


Based on work by Chase Stuart, I estimate a replacement-level veteran as providing 3.36 AV. Extrapolating Stuart's estimates of the value of salary cap dollars to the current salary cap, I have $1.587 million as the value of a one-unit increase in AV.

Let's start by looking at Malik Jackson. First, I adjust his contract for the amount of guaranteed money to come up with an adjusted average of $17.122 million. (Last year, I just used the annual average, but I wanted to improve on that, since the contract's value depends in large part on the amount of the guarantee. Olivier Vernon's contract, for example, provides more value than Jackson's despite a slightly lower annual average.) That number is close to the actual average of $17.1 million because Jackson's guarantee was about what you'd expect for a contract of that amount and length.

Jackson's adjusted average of $17.14 million is $16.38 million more than the veteran's minimum of $760,000. I estimate that Jackson thus needs to generate 13.68 units of AV to be worth his contract.

Needed Future AV = 3.36 + ($16.380 million)/($1.587 million) = 13.68

To project Jackson's performance for future years, I used a regression that looked at his performance over the last three years and his age. The regression accounts for potential nonlinearities in the relationship. From this projection, I get Age-Adjusted Value, my estimate of the value that a contract either creates or destroys. Malik Jackson's Age-Adjusted Value of -5.90 projects him to produce 5.90 AV less than he needs to in order to justify his contract.

A few notes:

1) After a suggestion last year, I checked out whether it made sense to include the idea that jumping from 19 AV to 20 AV is more valuable than jumping from 4 AV to 5 AV. In other words, even the marginal jump might be worth less at the lower end of performance than at the upper end. I checked this out a variety of ways and found no clear evidence for that idea. But I do think there is something to the idea that AV overvalues poor starters. For example, this year Chargers defensive end Kendall Reyes racked up 7 AV in large part just by starting 15 games. For this reason, I think the analysis works best outside of the bottom of free agency where some contracts that look like bargains are really just proper value for poor starters. As a result, I'm projecting only players who got at least $3,000,000 in average salary.

2) I've adjusted things across positions so the average running back contract is close to zero, the average cornerback contract is close to zero, etc. This is particularly important for running backs, who tend to be overvalued by AV.

3) I'm leaving out the quarterbacks since there aren't enough contracts at that position in the last two years to draw any meaningful conclusions, and they really need a separate model.

4) All contracts included are found in Spotrac's free agent contract database as of 11 p.m. Eastern on March 15. If anyone was either not included in those data because their contract terms were not yet included (e.g., Chris Long and Kelvin Beachum) or because they were not yet signed, they are excluded from Age-Adjusted Value calculations.

5) I've gone with the Spotrac data everywhere, both on guaranteed money (where they sometimes disagree with overthecap.com) and on player position.

The Good Deals

Player Pos Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Derrick Johnson ILB 33 KC 3 $7,000,000 $12,000,000 11.19 7.39 3.80
Marvin Jones WR 26 DET 5 $8,000,000 $20,000,000 10.84 7.94 2.90
Jermaine Kearse WR 26 SEA 3 $4,500,000 $6,300,000 9.19 5.74 2.80
Danny Trevathan ILB 25 CHI 4 $6,125,000 $12,000,000 9.52 6.76 2.76
Adam Jones CB 32 CIN 3 $6,666,667 - 9.35 6.70 2.65
Bobby Massie RT 26 CHI 3 $6,000,000 $6,500,000 9.15 6.59 2.56
Cedric Thornton DE 27 DAL 4 $4,250,000 $9,000,000 8.05 5.61 2.44
J.R. Sweezy G 26 TB 5 $6,500,000 $14,500,000 9.34 6.94 2.40
Brandon Mebane DT 31 SD 3 $4,500,000 $5,500,000 8.08 5.71 2.37
Jerrell Freeman ILB 29 CHI 3 $4,000,000 $6,000,000 7.80 5.45 2.35
Jaye Howard DT 27 KC 2 $5,000,000 $8,260,000 8.48 6.29 2.19
Rodney McLeod FS 25 PHI 5 $7,000,000 $17,000,000 9.37 7.29 2.08
Mike Wallace WR 29 BAL 2 $5,750,000 - 8.16 6.19 1.97
Matt Forte RB 30 NYJ 3 $4,000,000 $9,000,000 7.52 5.57 1.95
Vincent Rey ILB 28 CIN 3 $3,500,000 $3,000,000 6.86 5.04 1.82
Player Pos Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Tracy Porter CB 29 CHI 3 $4,000,000 $4,250,000 7.14 5.38 1.76
Nigel Bradham OLB 26 PHI 2 $3,500,000 $4,500,000 6.79 5.21 1.58
Coty Sensabaugh CB 27 LA 3 $4,666,667 $6,500,000 7.24 5.84 1.40
Damon Harrison DT 27 NYG 5 $9,250,000 $24,000,000 10.17 8.78 1.39
Mohamed Sanu WR 26 ATL 5 $6,500,000 $14,000,000 8.18 6.93 1.25
Tashaun Gipson FS 25 JAC 5 $7,200,000 $12,000,000 8.20 7.28 0.92
Ben Jones C 26 TEN 4 $4,375,000 $7,500,000 6.44 5.63 0.81
Brent Grimes CB 32 TB 2 $6,750,000 $7,000,000 7.98 7.20 0.78
Tahir Whitehead OLB 25 DET 2 $4,000,000 $4,750,000 6.06 5.50 0.56
Travis Benjamin WR 26 SD 4 $6,000,000 $13,000,000 7.07 6.72 0.35
Eugene Sims DE 29 LA 3 $3,333,333 $3,750,000 5.16 4.98 0.18
Akiem Hicks DE 26 CHI 2 $5,000,000 $5,500,000 6.27 6.11 0.16
Jermaine Gresham TE 27 ARI 1 $3,500,000 $1,000,000 5.12 5.06 0.06
Coby Fleener TE 27 NO 5 $7,200,000 $18,000,000 7.47 7.43 0.04
Phil Dawson K 41 SF 1 $3,134,000 $3,134,000 5.12 5.12 0.00

These are the major free agent contracts that Age-Adjusted Value rates as bringing value to the signing team. Thinking of this as one piece of the puzzle that should include what we see on film, a bunch of contracts jump out to me as particularly appealing, offering big upside to the signing team. The numbers love getting Jaye Howard at the price the Chiefs paid, and so do I. And Howie Roseman has had one incredible offseason. Not only has he obtained something in return for dumping contracts that have negative value, he has also made two of the best value signings in free agency, getting 25-year-old safety Rodney McLeod and 26-year-old linebacker Nigel Bradham at numbers that offer lots of upside.

The analysis may be a little tilted towards loving the Derrick Johnson deal since his second-team All-Pro honor gives a big boost in PFR's estimate of his performance. Age-Adjusted Value's five favorite contracts averaging over $7 million per season belong to Johnson, Marvin Jones, McLeod, Damon Harrison, and Tashaun Gipson. Yes, props to the Jags for the Gipson deal.

The Bad and Ugly Deals

Player Pos Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Lamar Miller RB 24 HOU 4 $6,500,000 $14,000,000 6.99 7.03 -0.04
Casey Hayward CB 26 SD 3 $5,100,000 $6,800,000 5.96 6.10 -0.14
Mitchell Schwartz RT 26 KC 5 $6,600,000 $20,700,000 6.95 7.16 -0.21
Ladarius Green TE 25 PIT 4 $5,000,000 $4,750,000 5.43 5.90 -0.47
Rishard Matthews WR 26 TEN 3 $5,000,000 $5,000,000 5.48 5.97 -0.49
Paul Soliai DT 32 CAR 2 $3,500,000 $3,000,000 4.37 5.11 -0.74
Derrick Shelby DE 27 ATL 4 $4,500,000 $7,500,000 4.93 5.70 -0.77
Jeff Allen G 26 HOU 4 $7,000,000 $12,000,000 6.32 7.25 -0.93
Sean Smith CB 28 OAK 4 $9,500,000 $20,000,000 8.01 8.97 -0.96
Al Woods DT 28 TEN 3 $3,500,000 $5,000,000 4.13 5.13 -1.00
Bruce Irvin OLB 29 OAK 4 $9,250,000 $19,000,000 7.74 8.79 -1.05
Isa Abdul-Quddus FS 27 MIA 3 $4,250,000 $6,000,000 4.41 5.59 -1.18
Ben Watson TE 35 BAL 2 $3,500,000 $3,000,000 3.84 5.11 -1.27
Brandon Brooks G 26 PHI 5 $8,000,000 $21,000,000 6.64 7.97 -1.33
Player Pos Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Alex Mack C 30 ATL 5 $9,000,000 $28,500,000 7.33 8.75 -1.42
Chris Ivory RB 27 JAC 5 $6,400,000 $10,000,000 5.11 6.77 -1.66
Donald Stephenson LT 27 DEN 3 $4,666,667 $10,000,000 4.14 5.99 -1.85
Eric Weddle FS 31 BAL 4 $6,500,000 $13,000,000 5.00 6.98 -1.98
Janoris Jenkins CB 27 NYG 5 $12,500,000 $28,800,000 8.84 10.85 -2.01
Jeremy Lane CB 25 SEA 4 $5,750,000 $11,000,000 4.47 6.51 -2.04
Tyvon Branch SS 29 ARI 2 $4,000,000 $5,000,000 3.45 5.52 -2.07
Nick Perry OLB 25 GB 1 $5,050,000 $1,500,000 3.72 5.99 -2.27
Marcedes Lewis TE 31 JAC 3 $4,000,000 $5,000,000 3.11 5.41 -2.30
Robert Ayers DE 30 TB 3 $6,500,000 $10,500,000 4.43 7.04 -2.61
Kelechi Osemele LT 26 OAK 5 $11,700,000 $25,400,000 7.15 10.29 -3.14
Bilal Powell RB 27 NYJ 3 $3,750,000 $6,000,000 1.80 5.31 -3.51
Olivier Vernon DE 25 NYG 5 $17,000,000 $52,500,000 9.00 13.90 -4.90
Malik Jackson DE 26 JAC 5 $17,100,000 $42,000,000 7.78 13.68 -5.90

A bunch of very good players appear on the list of the deals that leave the signing team worse off for signing them. At the right price, any team would be thrilled to add players such as Alex Mack or Eric Weddle. But Age-Adjusted Value finds the 30-year-old Mack not worth his big deal ($28.5 million guaranteed), nor 31-year-old Eric Weddle worth his ($13 million guaranteed).

The Jets' signing of Bilal Powell didn't move the needle much, but it was one of the most perplexing moves of free agency. Age-Adjusted Value ranks the contract for the eminently replaceable Powell as the third-worst value of free agency so far. (The Age-Adjusted Value would be somewhat higher if we replace Powell's projection with replacement-level play.)

But it's the Jets' co-tenants at MetLife Stadium who are so far challenging the Jaguars in failing to get a bang for their free-agent buck, even with the value in the Damon Harrison contract. The Giants paid Janoris Jenkins an average salary that is only $1.5 million less than what the Seahawks gave Richard Sherman last season. Age-Adjusted Value thinks that's too much for Jenkins even as he enters his prime seasons. And Age-Adjusted Value is almost as down on the Olivier Vernon contract as it is on the Malik Jackson deal. With the huge guarantee, the contract is effectively even richer than Jackson's. To live up to his deal, Vernon needs to become one of the five or six best defensive ends in football.

Posted by: Andrew Healy on 16 Mar 2016

23 comments, Last at 21 Mar 2016, 7:16pm by Andrew Healy


by Jimmy Oz :: Wed, 03/16/2016 - 5:53pm

AV ranks JR Sweezy as a better guard than Kelechi Osemele.

by Bowl Game Anomaly :: Wed, 03/16/2016 - 6:45pm

AV isn't designed to compare players in individual years, and especially not offensive linemen. It's intended to compare players over entire careers, and even then with a grain of salt.

by theslothook :: Wed, 03/16/2016 - 7:03pm

Yep, Av has a lot of problems. A LOT

by Jimmy Oz :: Wed, 03/16/2016 - 7:16pm


"Created by PFR founder Doug Drinen, the Approximate Value (AV) method is an attempt to put a single number on the seasonal value of a player at any position from any year" or to paraphrase, 'AV is designed to compare players in individual years'.

by njwelch75 :: Wed, 03/16/2016 - 8:51pm

You're not paraphrasing that very well. They assign a number to every player in every year so that they can be compiled later in terms of draft class, or career value. Even then, they admit there are issues with it (bad players may start on bad teams and good players may sit on good teams). You say that AV rates a player ahead of another (by 4 points over 4 years) and use that as a reason to entirely dismiss the stat. I think that's a mistake.

But yes AV has many issues, and using it to place a monetary value on players who have started as few as 4 seasons is not a good idea.

by Jimmy Oz :: Thu, 03/17/2016 - 12:38am

You have done me a disservice and misrepresented my argument. I don't use Sweezy Osemele as a reason to dismiss AV. I use it as the most egregious example in the above list of AV's failure to provide useful data. I have previously dismissed AV as nothing more than a curiosity.

by ishalev :: Wed, 03/16/2016 - 7:16pm

I haven't been part of the other discussions on this topic, but it's very similar to modeling values for fantasy auction leagues. You mention in the article that you don't see any evidence that going from AV 19 to 20 is better than 4 to 5. But have you considered starting player limits?

In fantasy, if you took total cap, and divided it by production over average (VBD) you'd get $/pts. But that models assumes you can score all those points, and you can't. You can't buy 5 RBs who are 1/5 as good as the top guys and start them all. In other words, concentration of production in a single person is good, b/c there's another limited resource to consider besides money, which is playing time. The better way to think about this is not AV, but AV/time played, or AV/snaps. The issue here is a rate equation not a compiling stat.

I'm not sure your model addresses this, but when you apply these ideas in the fantasy context, you get a value curve, where the marginal VBD point is has an increasing value.

by Andrew Healy :: Wed, 03/16/2016 - 9:47pm

To make a long story short, yes. I considered this and a bunch of other ways their could be nonlinearities in marginal jumps in AV. Basically, I've kept things simple because I'd say the evidence was inconclusive. But that's very different from saying I have a strong feeling that it's very nearly linear.

The weaknesses in AV include, but are not limited to, positions getting valued poorly (RBs too much, for example) and starters being overvalued (maybe the biggest single problem). The latter problem may mean that moving up on the low end of AV just means moving from bench players to starters on bad teams without being a real increase in quality. This is a potential reason why moving from 14 to 15 AV could be more valuable from 4 to 5 AV. Even if the marginal increase was equally valuable in a world where we measured performance perfectly, since we don't the former increase could be more likely to capture an actual increase in performance. But then some of the jumps at higher levels are about Pro Bowl votes, which have their own flaws. So many issues to consider...

Anyway, AV has its issues. But estimating value across positions is a really hard thing, and PFR's done a great service by making a good attempt at this despite lots and lots of difficulties. It's certainly true, though, that my Age-Adjusted Value numbers are just a rough guideline since they're based on a solid number (the needed AV) and a noisy one (the projected AV based on AV in recent years).

by mehllageman56 :: Wed, 03/16/2016 - 7:37pm

I don't see how the AV value for Damon Harrison's contract can be in the positive, unless something is really wrong with AV. He's a dominant run-stuffing nose tackle going to a 4-3 team whose Super Bowl runs have been based on playing edge rushers inside. He's getting 9 million a year, 6 million less than Wilkerson's getting for being franchised at end. Plus, they signed his back-up, Kendrick Ellis, last year, and then cut him midway through the year. It's not like Ellis was a bad player for the Jets, he just didn't fit the Giants' system.

by Thekdog :: Wed, 03/16/2016 - 8:16pm

Where would Osweiler rank on this?

by Bowl Game Anomaly :: Thu, 03/17/2016 - 9:14am

This kind of analysis isn't really relevant to Osweiler. If he's good, it's a good contract. If he's bad, it's a bad contract. There's no benchmark for him to hit that makes the contract a good value, because ultimately Houston will be judged on the mere fact of signing him, regardless of what they pay.

by bingo762 :: Thu, 03/17/2016 - 10:18am

Where would Zach Ertz extension put him on this list?

by garion333 :: Thu, 03/17/2016 - 11:14am

"No team came close to the Jaguars last year in expected value lost by overpaying for free agents."

Because it's the only way to bring people to Jacksonville. Perhaps they'll soon turn their fortunes around and be attractive in the way Oakland is beginning to look.

by Shaniya :: Thu, 03/17/2016 - 11:48am

Worst website ever! Can't wait to see it dying.

by tuluse :: Thu, 03/17/2016 - 11:57am

Can you add a link to last year's article so we can see how the projections have worked out so far?

My commentary on this article:

I think there needs to be a length of contract adjustment. Nick Perry might be overpaid, but it's for 1 year, so it hardly matters. As an example, you think Perry will under-perform his contract by 2.27 AV, but you think Eric Weddle will under-perform by 1.98 per year, on a 4 year contract. That's 7.92 total difference.

Also, I'll agree with the above posters that a linear value of AV is probably not the best model. It's still interesting info though.

by Andrew Healy :: Thu, 03/17/2016 - 2:06pm

Link is in the second paragraph above. Last year's worst value: Byron Maxwell.

And I agree on length of contract. The age adjustment gets at that a bit. But you have to decide whether adding that extra layer is worth the added complexity. Will think about it some more.

by tuluse :: Thu, 03/17/2016 - 3:29pm

Thanks for pointing that out. Missed the link.

I think you should have given McPhee the Osweiler treatment last year.

by ChrisS :: Thu, 03/17/2016 - 1:40pm

Does the annual increase in the salary cap effect these numbers in a material way? If the average salary gap goes up by 6% a year then the average divisor over a four year contract should be about $1.75m per year not $1.587m. From 2013 to 2016 the salary cap went from $123m to $155m about 6% compounded.

by medelste :: Fri, 03/18/2016 - 8:34am

Andrew, how about a little love for Chicago's Ryan Pace? 5 signings (including 3 of the top 10) on the Good Deals list, and no appearances on the Bad Deals list?

by Andrew Healy :: Mon, 03/21/2016 - 7:16pm

Wow, great point. For all the caveats about how these are just rough estimates, being in the plus zone on all five is a very good sign.

by ChicagoRaider :: Mon, 03/21/2016 - 7:07am

How "approximate" is AV? Is there an evaluation of the standard deviation of projections from it? It is hard to evaluate the assignment of "good" versus "bad" without knowing that. I somehow think that there is no way that second decimal place is significant, and the first one likely isn't either.

Even taking the numbers at face value, players play in a context. For example, the Raiders are in the strange position of having a lot to spend and needing to find players willing to take their money. Not all players are into the "this is a rising team" pitch. It looks like the signees extracted "you are still losers" money.

Players also play in units. Value to the team would be how much the player is worth to the particular team, not to the average team. So evaluating Osemele would be asking how much better is the Raider offense with Osemele, not how good Osemele is standing alone. They probably overpaid by that measure anyway, but I suspect by not as much as the chart suggests.

by Jerry :: Mon, 03/21/2016 - 7:58am

How "approximate" is AV?

Very. It's designed to put a number on every player every season. As Doug Drinen says on the page linked in #4, "If one player is a 16 and another is a 14, we can't be very confident that the 16AV player actually had a better season than the 14AV player. But I am pretty confident that the collection of all players with 16AV played better, as an entire group, than the collection of all players with 14AV."

It's designed more for aggregating seasons than for comparing individual seasons. To be fair, Andrew knows this. He wants something that will allow him to compare a guard, a running back, and a safety, and AV is better than the currently available alternatives.

by eggwasp :: Mon, 03/21/2016 - 8:49am

A couple of comments
1)the Raiders & Jags (for example) HAVE to spend the money or they are going to get caught by the Cap rules for underspending. For these teams, it doesn't matter if you overpay free agents - in fact, it may be beneficial to you given the cap rules. A team close to the cap gets effectively penalised far more from even a slight overspend (through lack of flexibility to do other things).

2) If a team has a glaring hole, it makes sense for them to spend more to fill it. If an O-Line is terrible all over, then overpaying for one quality lineman might be a waste of money - there are plenty of gaps for the defense to run through... OTOH, the Raiders had just one OL hole to fill - doesn't it make sense to upgrade significantly (at higher expense) to turn it into an elite unit (hopefully...)?