2017 Free Agency Cost-Benefit Analysis

2017 Free Agency Cost-Benefit Analysis
2017 Free Agency Cost-Benefit Analysis
Photo: USA Today Sports Images

by Bryan Knowles

You can almost set your watch by the annual shock over major NFL contracts. With the salary cap increasing by more than $10 million for the fourth consecutive season, we have seen yet another round of eye-popping or record-setting deals. $67.5 million for a cornerback with one year of starting experience! A record $17 million a year for a wide receiver! $45 million for something called a "Mike Glennon!"

Some teams and fanbases see free agency as a "get out of purgatory free" card. The Jacksonville Jaguars, for example, have already added $174.4 million in contracts this offseason, including $69.6 million in guaranteed money, in an attempt to get back to relevance. This is the third season in a row they have been in the top two in free-agent spending, which is perhaps the best piece of evidence that you can't simply buy your way to success in the NFL.

No, to succeed in free agency, you want to be the team with the most cost-effective deals. You don't win championships by overpaying big-name stars. You win by having players outperform their contracts and getting more value out of your acquisitions than your opposition. In other words, you want your free-agency deals to be more Drew Brees in 2006 than Brock Osweiler in 2016.

Obviously, the only way to really judge a free-agent class is to look back and see how players performed with the benefit of hindsight. General managers don't have that luxury when signing players, however. So, in the interest of judging process rather than results, this our third annual attempt to identify the best and worst values of the free-agent market.


We have made some tweaks to this year's model in an attempt to better represent the actual realities of the free-agent market. If you don't care how the sausage is made and just want to see the results, you can safely skip this section.

Once again, we're using Pro Football Reference's Approximate Value statistic as our measure of player quality. As you might expect from a stat that seeks to quantify a wide range of different positions and skillsets across decades of NFL play and squash it down into a single number, it's not without its issues. It can't really differentiate between a great blocking tight end and a bad receiving tight end, gives too much value to poor players who nevertheless manage to stay in the starting lineup, and has serious trouble assigning appropriate values to individual offensive linemen (particularly bad linemen on good teams and vice versa).That being said, it's still a very useful tool for comparing a wide variety of players quickly and effectively. As long as we keep in mind that it's called approximate value and don't fret too much about the difference between an 8 AV player and a 7 AV player, it'll do nicely for our analysis.

In theory, a point of AV is a point of AV. The 16 AV put up by both Khalil Mack and Alex Mack in 2016 is equivalent, across positions. In previous years, we have used that to ascertain how much value a player needed to produce to justify their contract, but it turns out, position does matter here, and using one salary baseline for every position isn't the best option.

For example, once you remove rookies, minimum contracts and replacement-level players from the mix, teams spend about 1.5 times as much per AV on quarterbacks as they do on running backs. Are front offices overvaluing quarterbacks? Is the demand for quality quarterback play greater than the demand for rushers? Is AV overrating the importance of running backs? Yes, to all of the above. The end result is that it makes more sense to adjust the value of each contract for its position, rather than make a broad claim that all quarterbacks are overpaid.

Based on work by Chase Stuart, we can estimate a replacement-level veteran as providing 3.36 AV. We want to determine how much extra value a player needs to add in order to justify his contract. That value varies from position to position, with the league average being approximately $1.7 million per point of AV added.

Take Dont'a Hightower, for example. Hightower's new contract with New England is for four years and $35.5 million, or $8.875 million per season. That's $8.1 million more than the veteran's minimum of $775,000, so he needs to generate $8.1 million worth of value above replacement each season for his new deal to be worth it.

Linebacker is generally one of the cheaper positions to fill. Looking at veteran contracts over the past three seasons and adjusting for the expanding salary cap, we can determine that an average veteran linebacker costs about $1.13 million per point of AV added (in 2017 cap dollars). Now, we can estimate Hightower's needed AV to justify his contract:

Needed Future AV = 3.36 + ($8.1 million)/($1.13 million) = 10.53

To project Hightower's performance for future years, we 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, we get Age-Adjusted Value, our estimate of the value that a contract either creates or destroys. Don’t'a Hightower's Age-Adjusted Value of plus-0.03 projects him to produce 0.03 AV more per season than he needs to in order to justify his contract.

Some final notes before we get to the numbers:

1) We're limiting the analysis to players with at least $3 million in average salary. AV tends to overvalue poor starters, letting them rack up AV for simply staying in the lineup. By setting a floor for the projections, we ignore some cases at the bottom of free agency where contracts that look like bargains are actually just proper value for low-quality starters.

2) We're using Spotrac's list of signed unrestricted free agents for all of our contract details. No restricted free agents, contract extensions, or other new 2017 contracts were included in the analysis was included. All information was correct as of March 24.

3) Age-adjusted value is a per-season metric. Obviously, some contracts last longer than one season, and a bad deal on a one-year contract hurts far less than a bad deal on a five-year contract. I'll point out particularly egregious long-term deals as we go.

Have I lost you yet? Excellent! Let's get to the actual numbers.

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Brian Hoyer 31 SF 2 $6,000,000 $6,950,000 7.93 5.47 2.46
Nick Foles 28 PHI 2 $5,500,000 $7,000,000 5.90 5.23 0.67
Josh McCown 37 NYJ 1 $6,000,000 $6,000,000 3.54 5.47 -1.93
Mike Glennon 27 CHI 3 $15,000,000 $18,500,000 4.90 9.72 -4.82

How do you project someone like Mike Glennon? Glennon has attempted 11 passes in the last two years. Take those away and we are left with a handful of games from three seasons ago and some preseason performances. For what it's worth, Glennon was 24th in DVOA during his six games in 2014; $15 million a season puts him 22nd in salary among NFL quarterbacks. Chicago is putting a lot of faith in its scouting to give Glennon that much money, even if it can get out of the deal after just one year if everything goes south.

The Bears likely could have kept Brian Hoyer. The 31-year-old Hoyer is no one's quarterback of the future, but he is one of the NFL's better backup or bridge options. At $6 million a year, Hoyer will rank 27th among NFL quarterbacks. That would be a massive steal for someone who reached 19.5% DVOA last season, and is still a good value for a player who was at -3.0% and -5.3% in two years as a starter before that. Reuniting with Kyle Shanahan in San Francisco, Hoyer could lead the 49ers to the dizzying heights of a sixth win in 2017.

Running Backs

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Eddie Lacy 26 SEA 1 $4,250,000 $2,865,000 7.32 5.65 1.67
Rex Burkhead 26 NE 1 $3,150,000 $1,100,000 2.85 4.81 -1.96
Kyle Juszczyk 25 SF 4 $5,250,000 $7,000,000 2.83 6.41 -3.58

Remember when Eddie Lacy wowed the world in his first two seasons, averaging 1,500 yards from scrimmage, making the Pro Bowl, and looking like a top-10 running back for years to come? Weight issues and injuries have kept him quiet since then, but he's still only 26 -- plenty of time for a career comeback. If he could somehow regain his 2014 form, Seattle's one-year deal would be an amazing value. That would require him to be both healthy and in shape, however, and Lacy reportedly tipped the scales at 267 pounds during his free agency visits. Color me a little skeptical here.

Kyle Juszczyk is now the highest-paid fullback in football, making more than twice what Ryan Hewitt makes. It's a staggering amount of money, even for arguably the best fullback in the game. Kyle Shanahan used Patrick DiMarco to great effect in Atlanta, and AV isn't properly giving Juszczyk credit for his excellent blocking, so this deal might not be quite as bad as the numbers indicate. Still -- DeMarco is getting just $2.1 million per season in Buffalo; is Juszczyk really worth two and half times as much?

Wide Receivers

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Terrance Williams 27 DAL 4 $4,250,000 $9,500,000 5.76 4.72 1.04
Ted Ginn Jr. 31 NO 3 $3,666,667 $3,000,000 5.44 4.43 1.01
Brandon Marshall 32 NYG 2 $5,500,000 $5,000,000 6.16 5.34 0.82
Terrelle Pryor 27 WAS 1 $6,000,000 $6,000,000 6.30 5.59 0.71
Robert Woods 24 LARM 5 $6,800,000 $7,000,000 6.48 5.99 0.49
Cordarrelle Patterson 26 OAK 2 $4,250,000 $5,000,000 4.84 4.72 0.12
Markus Wheaton 26 CHI 2 $5,500,000 $6,000,000 5.22 5.34 -0.12
Marquise Goodwin 26 SF 2 $3,000,000 $4,450,000 3.90 4.10 -0.20
Pierre Garcon 30 SF 5 $9,500,000 $17,000,000 6.87 7.33 -0.46
DeSean Jackson 30 TB 3 $11,166,667 $20,000,000 7.04 8.16 -1.12
Alshon Jeffery 27 PHI 1 $9,500,000 $8,750,000 6.16 7.33 -1.17
Kenny Britt 28 CLE 4 $8,125,000 $10,500,000 5.46 6.65 -1.19
Torrey Smith 28 PHI 3 $5,000,000 $500,000 3.63 5.10 -1.47
Russell Shepard 26 CAR 3 $3,333,333 $2,100,000 2.64 4.27 -1.63

Most people would peg Terrance Williams as a low-end No. 2 wideout. DVOA doesn't exactly agree; though Williams struggled for targets behind Cole Beasley, Dez Bryant, and Jason Witten in 2016, he finished fourth in DVOA when the ball was thrown his way. Still, at $4.25 million a season, he's now the 40th-highest paid receiver in football; I'd subjectively have that as a minor overpay, if anything.

AV has no way of knowing that Ted Ginn is apparently allergic to catching passes from anyone not named Cam Newton; Ginn has averaged about a reception and 20 yards more per game in Carolina blue than he has anywhere else. If he can build some chemistry with Drew Brees, however, Ginn might slide into Brandin Cooks' old deep-threat role. $3.7 million a season seems like a fairly decent deal for a situational deep threat, even for a 32-year-old Ginn.

[ad placeholder 3]

DeSean Jackson is projected as the best receiver in this bunch, but nearly $12 million a season for a receiver past his 30th birthday is a very large pill to swallow. As long as Jackson keeps putting up 1,000 yard seasons, the contract is fine. However, it's tough for receivers to keep up that high level of performance into their 30s. The size of the deal means the Buccaneers are realistically tied to Jackson for the next two seasons, paying him like a top-10 receiver. He arguably was a top 10 receiver last season, but if his speed dries up in the next few years, the deal may well feel like an albatross.

But at least adding Jackson to give Jameis Winston a deep threat makes sense on paper. Cleveland's wide receiver shuffle makes less sense. They had a chance to keep Terrelle Pryor, who ended up on a short-term deal in Washington. AV's likely underrating Pryor here, because there's really no way to take into account the fact that last season was essentially his first as a receiver.

Rather than reap the benefits of the time and effort they spent working on Pryor's development, however, Cleveland gave a four-year deal to Kenny Britt. True, Britt had a slightly higher DVOA than Pryor last year in an even worse offense. But Britt is older, more expensive even after Pryor's incentives are taken into account, and less likely to improve in the future. Statistics suggest he may even be a worse receiver right now, While the one-dimensional Torrey Smith and the replacement-level Russell Shepard grade out as worse deals in the metric, they're essentially one-year deals; Cleveland is tied to Britt for at least two.

Tight Ends

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Martellus Bennett 30 GB 3 $7,000,000 $6,300,000 5.68 5.37 0.31
Jared Cook 29 OAK 2 $5,300,000 $5,000,000 3.69 4.72 -1.03
Ryan Griffin 27 HOU 3 $3,000,000 $3,225,000 2.78 3.85 -1.07
Mychal Rivera 26 JAC 2 $3,375,000 $750,000 2.41 3.99 -1.58
Levine Toilolo 25 ATL 3 $4,000,000 $3,000,000 2.51 4.23 -1.72
Dion Sims 26 CHI 3 $6,000,000 $6,000,000 2.30 4.99 -2.69
Rhett Ellison 28 NYG 4 $4,500,000 $8,000,000 0.66 4.42 -3.76

One of the most noted problems with AV is that it undervalues good blocking tight ends and receivers, and that's part of what you're seeing here, with so many tight ends being given significant negative grades. Still, it does feel, even subjectively, like a lot of relatively minor tight ends were given hefty deals in 2017, especially considering the deep class of tight ends in this year's draft. $6 million a year for Dion Sims? I'd have a hard time justifying that.

Ironically, considering these comments he made on NFL Network, the one tight end who isn't overpaid is Martellus Bennett. Bennett's not just a product of New England's offense; he put up good numbers in both Chicago and New York before getting to play with Tom Brady in New England. I also seem to remember that Green Bay has a pretty decent quarterback as well. Bennett should do quite well in Jared Cook's old role in Green Bay, even at age 30.

Offensive Line

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
D.J. Fluker 26 NYG 1 $3,000,000 $1,500,000 7.94 4.48 3.46
Andre Smith 30 CIN 1 $3,250,000 $1,550,000 6.37 4.65 1.72
Mike Remmers 27 MIN 5 $6,000,000 $10,500,000 8.16 6.48 1.68
Andrew Whitworth 35 LARM 3 $11,250,000 $15,000,000 10.33 9.98 0.35
Larry Warford 25 NO 4 $8,500,000 $10,100,000 7.75 8.14 -0.39
T.J. Lang 29 DET 3 $9,500,000 $19,000,000 8.24 8.81 -0.57
Kelvin Beachum 27 NYJ 3 $8,000,000 $12,000,000 7.24 7.81 -0.57
Luke Joeckel 25 SEA 1 $8,000,000 $7,000,000 7.03 7.81 -0.78
Benjamin Ijalana 27 NYJ 2 $5,500,000 $0 5.19 6.15 -0.96
Ricky Wagner 27 DET 5 $9,500,000 $17,500,000 7.11 8.81 -1.70
J.C. Tretter 26 CLE 3 $5,583,333 $6,500,000 4.36 6.20 -1.84
Menelik Watson 28 DEN 3 $6,125,000 $5,500,000 4.62 6.56 -1.94
Riley Reiff 28 MIN 5 $11,750,000 $26,300,000 8.07 10.31 -2.24
Matt Kalil 27 CAR 5 $11,100,000 $13,000,000 7.50 9.88 -2.38
Ron Leary 27 DEN 4 $9,000,000 $18,650,000 5.86 8.48 -2.62
Kevin Zeitler 27 CLE 5 $12,000,000 $23,000,000 7.29 10.48 -3.19
Russell Okung 29 LACH 4 $13,250,000 $25,000,000 7.51 11.31 -3.80

This is probably the position where we should take this analysis with the most grains of salt. AV undervalues linemen who were good but not good enough to make the Pro Bowl, and overvalues bad offensive tackles who get a lot of starts. That tends to work itself out over enough time and players, but when you're dealing with smaller sample sizes, outliers do tend to pop out.

That explains, for example, why Mike Remmers is so high. Remmers started 37 games at right tackle over the past two seasons for a team that went to the Super Bowl; he has to be pretty good right? Well, no. Sports Info Solutions charted Remmers with the fourth-most missed blocks in the league in 2016, and our own Ben Muth called him "probably the worst full-time starter" he watched. Who am I to argue offensive line play with Ben Muth? Remmers is getting the benefit of being associated with a good team, as opposed to being valued for his own skill level here.

Left tackles also seem to be a bit depressed overall -- they probably need to have a separate contract modifier from right tackles, and that's a project to (ahem) tackle for future editions. That being said, it's impressive that, even at age 35, Andrew Whitworth is still outperforming major free-agent contracts. The formula is slamming him with age-related penalties as most offensive linemen just don't continue to play at all through their mid-30s, much less make a All-Pro squads. Jared Goff should thank his lucky stars Whitworth's coming to town.

I may not be as high on the D.J. Fluker deal as the stats indicate, but a one-year, $3 million deal is very low risk for someone who has played well when healthy and can cover both tackle and guard spots.

Russell Okung is a major injury question mark. While he played all 16 games last year, he missed 13 in the three seasons prior. That makes a contract with $25 million guaranteed quite the gamble for the Los Angeles Chargers; Okung is now the second-highest paid offensive lineman in football, behind Trent Williams, and that's a bit of a gamble. Similarly, the metric doesn't like that Kevin Zeitler is now the highest-paid guard; it's hard to be a value when you make more money than anyone else at your position.

Subjectively, though, I'd pin Carolina signing Matt Kalil as the worst of the bunch so far. Not only is Kalil coming off of a major hip surgery, but he also really hasn't had a particularly solid season since 2012. With the contract structure being what it is, they probably can't slip out of it particularly easily until after the 2020 season, so it's a heck of a gamble, and not one I would be comfortable making.

Defensive Line

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Chris Baker 29 TB 3 $5,250,000 $6,000,000 8.01 6.27 1.74
Domata Peko 32 DEN 2 $3,750,000 $3,800,000 6.67 5.18 1.49
Lawrence Guy 27 NE 4 $3,350,000 $4,900,000 6.31 4.88 1.43
Tyson Alualu 29 PIT 2 $3,000,000 $1,750,000 5.41 4.63 0.78
Sylvester Williams 28 TEN 3 $5,833,333 $7,000,000 7.45 6.70 0.75
Akeem Spence 25 DET 3 $3,000,000 $3,500,000 4.97 4.63 0.34
Terrell McClain 28 WAS 4 $5,250,000 $10,500,000 6.34 6.27 0.07
Dontari Poe 26 ATL 1 $8,000,000 $7,500,000 8.30 8.29 0.01
Ricky Jean-Francois 30 GB 1 $3,000,000 $0 4.38 4.63 -0.25
Stacy McGee 27 WAS 5 $5,000,000 $9,000,000 5.02 6.09 -1.07
Earl Mitchell 29 SF 4 $4,000,000 $4,650,000 4.18 5.36 -1.18
Calais Campbell 30 JAC 4 $15,000,000 $30,000,000 12.03 13.41 -1.38
Nick Fairley 29 NO 4 $7,000,000 $9,000,000 6.12 7.55 -1.43
Bennie Logan 27 KC 1 $8,000,000 $7,680,000 6.76 8.29 -1.53
Brandon Williams 28 BAL 5 $10,500,000 $24,500,000 7.93 10.11 -2.18

Calais Campbell is the best player on this list, but it's likely that the Jaguars are overpaying for past production on this deal. As long as Campbell keeps playing like has over the past few seasons in Arizona, the Jaguars are getting one of the best 3-4 defensive ends in football. The Jaguars don't actually run a 3-4, mind you, but Campbell should be fine as an end in Todd Wash's system. The problem is that Campbell turns 31 in September. $15 million a year is a lot of money; if Campbell struggles with the transition to a 4-3 or begins to see his skills decline, that's a potential issue. Huge pick-up, big help for Jacksonville's defense -- but a big, big contract to take on.

Chris Baker is a fantastic signing; one of my favorite of the entire offseason. The Buccaneers have been looking for someone to stick alongside Gerald McCoy, and Baker's a great addition at a very reasonable price. He'll help make Tampa Bay a nightmare to run against, and he's not someone who needs to be replaced on passing downs. A very nice bit of value, there.

Some numbers here I definitely disagree with, however. Brandon Williams is one of the top interior linemen in the league, and he's still well within his prime. $10.5 million a year is a hefty price tag for sure, making him the NFL's highest-paid nose tackle, but it's nearly not bad enough to have him at the bottom of the table for me. On the other hand, Domata Peko's play hasn't nearly lived up to his actual AV scores. He's getting credit for remaining a starter on Cincinnati's defensive line, despite the fact that he hasn't really looked sharp since... 2012? Earlier? AV is definitely overrating him here.


Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Connor Barwin 30 LARM 1 $6,500,000 $0 6.96 6.39 0.57
Julius Peppers 37 CAR 1 $3,500,000 $1,650,000 5.10 4.61 0.49
John Simon 26 IND 3 $4,500,000 $5,500,000 4.19 5.20 -1.01
Datone Jones 26 MIN 1 $3,750,000 $1,600,000 3.65 4.76 -1.11
Jabaal Sheard 27 IND 3 $8,500,000 $9,500,000 6.04 7.57 -1.53

It should be noted that neither AV nor the Spotrac salary database has a distinction for edge-rushers, so splitting them out is a bit arbitrary. Comparing 3-4 outside linebackers to 4-3 outside linebackers has never really made sense, though, so it's a distinction that's worth making.

I'm really, really high on Connor Barwin for the upcoming season. Miscast as a 4-3 defensive end in Philadelphia last season, he gets to kick back out to 3-4 linebacker in Los Angeles. He also gets to work with Wade Phillips, who has magic defensive turnaround powers. I'd be shocked if he has five or fewer sacks again this season, barring injury.

Jabaal Sheard is probably best suited for a rotational pass-rushing role. He was benched late in the season in New England as Trey Flowers picked up steam, and looked more comfortable with some of the pressure taken off. You shouldn't really count on him to be a dominant force; he's a piece of the puzzle, and not the full answer. This all makes $8.5 million a year very questionable from Indianapolis. That's full-time starter money.


Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Kevin Minter 26 CIN 1 $4,250,000 $2,100,000 7.87 6.23 1.64
Lawrence Timmons 30 MIA 2 $6,000,000 $11,000,000 8.44 7.86 0.58
Malcolm Smith 27 SF 5 $5,300,000 $11,500,000 7.70 7.21 0.49
Dont'a Hightower 27 NE 4 $8,875,000 $17,000,000 10.56 10.53 0.03
Keenan Robinson 27 NYG 1 $3,000,000 $2,000,000 4.89 5.07 -0.18
Paul Worrilow 26 DET 1 $3,000,000 $2,750,000 4.77 5.07 -0.30
A.J. Klein 25 NO 4 $6,000,000 $9,400,000 4.38 7.86 -3.48

2016 was really a breakout season for Kevin Minter; the former second-round pick struggled to see the field in his first few seasons, but it seemed like the proverbial light went on for him last season. He's trending upwards, but you don't want to pay too much for one season of production. Cincinnati limits their risk with a one-year, $4.5 million contract.

A.J. Klein has been the weak link in the Panthers' corps over the past four seasons, and had his worst season in extended time last year, replacing Luke Kuechly after his concussion. Klein is alright as a run stopper, but asking him to cover tight ends or running backs is beyond his skill set. $6 million a year from the Saints is a hefty chunk of change; it makes him the 12th-highest paid inside linebacker in football. That seems like an awful lot for Carolina's fourth-best linebacker.

Defensive Backs

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Johnathan Cyprien 26 TEN 4 $6,250,000 $9,000,000 6.58 5.36 1.22
D.J. Swearinger 25 WAS 3 $4,500,000 $6,000,000 5.48 4.61 0.87
Antoine Bethea 32 ARI 3 $4,250,000 $4,000,000 5.35 4.50 0.85
Nolan Carroll 30 DAL 3 $3,333,333 $3,000,000 4.83 4.10 0.73
J.J. Wilcox 28 TB 2 $3,125,000 $3,125,000 4.70 4.01 0.69
Quintin Demps 31 CHI 3 $4,500,000 $4,500,000 5.23 4.61 0.62
Brandon Carr 30 BAL 4 $5,875,000 $4,000,000 5.54 5.20 0.34
Marcus Cooper 27 CHI 3 $5,333,333 $6,000,000 5.22 4.96 0.26
Terence Newman 38 MIN 1 $3,250,000 $1,500,000 3.99 4.07 -0.08
Tony Jefferson 25 BAL 4 $8,500,000 $19,000,000 6.09 6.33 -0.24
Barry Church 29 JAC 4 $6,500,000 $12,000,000 5.14 5.47 -0.33
Micah Hyde 26 BUF 5 $6,100,000 $10,300,000 4.90 5.29 -0.39
Captain Munnerlyn 28 CAR 4 $4,250,000 $8,800,000 4.08 4.50 -0.42
Jordan Poyer 25 BUF 4 $3,250,000 $6,000,000 3.62 4.07 -0.45
Darius Butler 31 IND 1 $3,000,000 $2,500,000 3.36 3.96 -0.60
Morris Claiborne 27 NYJ 1 $5,000,000 $2,000,000 4.11 4.82 -0.71
D.J. Hayden 26 DET 1 $3,750,000 $2,250,000 3.57 4.28 -0.71
Prince Amukamara 27 CHI 1 $7,000,000 $7,000,000 4.93 5.68 -0.75
Nate Allen 29 MIA 1 $3,400,000 $3,400,000 3.09 4.13 -1.04
Logan Ryan 26 TEN 3 $10,000,000 $12,000,000 5.93 6.98 -1.05
Stephon Gilmore 26 NE 5 $13,000,000 $31,000,000 7.11 8.27 -1.16
Kayvon Webster 26 LARM 2 $3,875,000 $4,250,000 2.72 4.34 -1.62
A.J. Bouye 25 JAC 5 $13,500,000 $26,000,000 5.14 8.48 -3.34

Age-Adjusted AV hates the A.J. Bouye contract, and it's not too difficult to see why. Bouye had only started eight games before 2016, and while he racked up a fair amount of interceptions, the lack of time in the starting lineup penalizes his AV score. It doesn't think Bouye was the top corner available, giving that nod to Stephon Gilmore. AV is wrong, of course; Bouye was amazing last year. Now, you can argue that the Jaguars are paying a lot of money on a long-term deal for a player who essentially has one year of good play in his history. Making him one of the five highest-paid corners in the game after starting last season fourth on the depth chart is a heck of a jump. Still, it's nowhere near as bad of a deal as the stats would indicate.

[ad placeholder 4]

Stephon Gilmore is a rare case of New England coming out looking bad in one of these stats. The Patriots generally do not make big deals like this one. Gilmore's charting stats have never been super-great; SIS had him as 77th out of 84 qualifying cornerbacks in yards per target last year, which isn't exactly great, making the decision to splurge on Gilmore in particular somewhat eyebrow-raising. (Gilmore did rank 25th the year before, but that was his career-best.) The Patriots had no negative contracts in either of the first two years of this analysis, but have two this year in Gilmore and Rex Burkhead, and a third just barely positive deal in Dont'a Hightower. This could be a sign that they're shifting from perpetual renewable value to an all-in run at a final championship or two under the Brady/Belichick combo, or they could simply have different evaluations of these players. By now, we have learned that second-guessing Belichick is a good way to look foolish.

On the other end of the scale, Johnathan Cyprien looks like a good signing for Tennessee. He's a hard-tackling, hard-hitting thumper, and it's likely he'll end up playing some linebacker for Dick LeBeau in dime packages. However, Cyprien had a career year just in time for him to cash in, so you have to worry about whether his performance is repeatable. You also don't want to count on him for a lot of pass coverage. Nonetheless, he'll only be 27 in 2017, and should be a great fit for the Titans.

Washington's offseason has been, shall we say, chaotic so far, but the signing of D.J. Swearinger is a high point. Swearinger performed well in Arizona's secondary last season, taking a significant step forward in pass defense. Also, it turns out he's just 25 and not 27 -- Elias had an odd mistake in their system, listing him with the wrong age. There's a big difference between 25 and 27! At 25, a safety is just entering his peak; at 27, he is at the age where he begins to decline. Age Swearinger by two years, and his projected AV would drop to 4.89.

Special Teams

Player Age Tm Yrs Average Guaranteed Proj AV Needed AV Age-Adj Val
Phil Dawson 42 ARI 2 $3,000,000 $1,500,000 3.60 3.40 0.20

For completeness' sake, here's the one sizeable special teams free-agent contract signed this offseason. Most kickers cost significantly less than $3 million per season. If you are going to spend that much on a kicker, Phil Dawson is a decent one to go grab; he has made 86.1 percent of his field goals over the last four seasons in San Francisco, though he is somewhat below par at kickoffs.

Conclusions and Final Thoughts

This isn't the end-all, be-all of free-agent analysis, but it gives us a good overview of how teams are choosing to spend their money. Even if our value estimations and projected future performance were 100 percent accurate -- which they are very much not -- that doesn't mean that a contract with a bad score is necessarily a bad decision. Seeing where teams deviate from these value calculations can give you an insight into how they're choosing to compete.

Some teams are gambling that their scouting or coaching can get more value out of a player than the general consensus would imply, like the Bears with Mike Glennon. Some are willing to pay an extra premium for the very best at a given position, like the Jaguars with Calais Campbell. Some have specific styles of play they want to highlight, like the 49ers with Kyle Juszczyk. Some are willing to overpay in order to plug the one hole that lies between them and contention, like the Vikings with Riley Reiff. Some are willing to gamble that one year of success becomes a trend, like the Jaguars (again) and A.J. Bouye. Some have earned enough value in other places to warrant a gamble, like the Patriots and Stephon Gilmore. And some teams simply have so much cap space that they can "afford" to overspend in order to attract free agents to less enticing situations, like the Jaguars (again) and the 49ers.

You're not going to build a long-term contender by filling your team with poor contracts. Simply signing the biggest names to the biggest deals very rarely ends up paying off. Competing in the NFL in the salary cap era is all about getting the most value out of your dollar, whether that's through finding that budget free agent which fits your scheme perfectly, extending a player who is about to take the next step under your coaching, or taking advantage of cheaper rookie contracts to fill out your lineup. It's alright to take a risk on a questionable contract every now and again, but the very best teams, the teams that contend year after year, don't make a habit of it. Stick to the Flukers and Bakers and pass on the Glennons and the Sims and you're more likely to be contending in January.

Special thanks go to Andrew Healy for the original cost-benefit analysis work over the last two seasons, to Chase Stuart for the initial calculations of replacement-level value, and to Doug Drinen for creating AV to begin with.


37 comments, Last at 15 Apr 2017, 8:32pm

1 Re: 2017 Free Agency Cost-Benefit Analysis

I appreciate all of the hard work. AV is a fine metric when we have goodish metrics for evaluating on field performance. So called "skill positions" work mostly. Edge rushers "kind of". The rest - who knows right?

I remain fascinated by cornerback play and how wildly it can swing year to year. This year, for instance, Richard Sherman had a mediocre season. Why did he have a mediocre season(And this apparently occurred even before Thomas' injury). Why did Chancellor do well this year when he was down last year?

Why would it not surprise me in the least to see Gilmore do well next year?

2 Re: 2017 Free Agency Cost-Benefit Analysis

In my opinion, AV works over the long run for the non-skill position players because it makes assumptions that front offices and coaches have some idea of what they're doing. It figures bad players won't get continued opportunities to prove themselves, and keep racking up games played and starts and things of that nature -- bad players get benched, bad players get cut. They stop racking up AV, and the good players rise to the top.

That works when looking over the course of a career just fine, but of course, breaks down on the individual season level. We've all seen situations where a terrible player keeps getting starts because the players behind them are worse, or because of massive injury situations. We've seen coaches fall in love with one of their draft picks and continue to try to develop them despite continued poor play on the field. That's where AV goes skewey; it can't really differentiate a starter based on merit and a starter based on the fact that ~someone~ has to start. It's trusting coaches to make that decision for them, and that doesn't always happen.

The three-year period DOES help quite a bit with that, but sometimes, bad situations last for more than three years (see Carolina's offensive tackles). Just good to look at context when judging those kinds of things, I suppose.

3 Re: 2017 Free Agency Cost-Benefit Analysis

I agree. I think in this context, when doing positional comparisons - AV works quite well. I've made this point but I'm not a huge fan of AV when it comes to draft analysis because it has a hard time distinguishing between returns by position. How much draft surplus is Adrian Peterson worth vs Orlando Pace?

4 Re: 2017 Free Agency Cost-Benefit Analysis

As a Seahawks fan, I did not interpret Sherman as having a "mediocre season."

I don't think cornerback performance does swing wildly from year to year. I think perception of cornerback performance swings wildly from year to year, and the kind of comprehensive, every snap, all-22, accounting for competition, context and scheme statistics which would prove this are all but impossible to record.

Week by week DVOA splits like vs. #1 Receivers or even Pass Defense are not available to me, but as of Week 12 before Earl Thomas was injured, the Seahawks had the fourth best overall defense.

I am not accusing you or whoever thinks Sherman had a mediocre season of this, but it seems fans often confuse the player closest to the free receiver for being the player who is responsible for that receiver being free. To know who's at fault, you must at minimum identify the scheme and responsibilities, which--for me anyway--is a darn time intensive task. Sherman is ultra aware, reads the quarterback pretty much every snap, is a good open-field tackler, and is among the quicker members of the Seahawks defense (whatever the reputation, it's a slow defense) and so he's often stuck chasing the free receiver.

Just as a fan who tends to watch every game a few times, I thought Sherman was very good. That scheme just can't survive the downgrade from Earl Thomas to Steven Terrell. It would take me thousands of words and diagrams to explain, but to borrow an idea from basketball, Thomas is the designed help defender on so many plays. Tho very fast Terrell's effective range was minuscule. What should have been converging double coverage or gang tackles became single coverage deep and one man to beat.

5 Re: 2017 Free Agency Cost-Benefit Analysis

I am referring to charting stats showing cornerback performance swings wildly year to year. As for Sherman - I was referring to his pff rating this year, though in fairness, they said he improved later in the year.

On a separate topic - I agree that Seattle's defense drops precipitously when Thomas is out. It makes me further marvel at NE which seems to be resistant to collapse when any one of their great players is done for the year, including Tom Brady.

7 Re: 2017 Free Agency Cost-Benefit Analysis

I originally had that in a draft of the article, but thought it was unfair, because it doesn't take into account renegotiations and restructures of their own contracts, or signing a number of cheaper,

That being said, here's the total value for each team:

1. Cincinnati: +3.36 (+1.68 per player)
2. Dallas: +1.76 (+0.88)
3. Tampa Bay: +1.31 (+0.44)
4. Arizona: +1.05 (+0.53)
5. Tennessee: +0.92 (+0.31)
6. Seattle: +0.89 (+0.44)
7. Pittsburgh: +0.78 (+0.78)
8. Washington: +0.58 (+.14)
9. NY Giants: +0.34 (+0.08)
10. Washington: +0.06 (+0.03)
11. LA Rams: -0.20 (-0.05)
12. Miami: -0.46 (-0.23)
13. Buffalo: -0.84 (-0.42)
14. Oakland: -0.92 (-0.46)
15. Houston: -1.07 (-1.07)
16. Kansas City: -1.53 (-1.53)
17. New England: -1.67 (-0.42)
18. Atlanta: -1.70 (-.0.85)
19. Minnesota: -1.75 (-0.44)
20. Philadelphia: -1.97 (-0.66)
21. Baltimore: -2.08 (-0.69)
22. San Francisco: -2.47 (-0.41)
23. Detroit: -2.94 (-0.59)
24. Denver: -3.07 (-1.02)
25. Indianapolis: -3.14 (-1.05)
26. LA Chargers: -3.80 (-0.98)
27. Carolina: -3.94 (-0.98)
28. NY Jets: -4.17 (-1.04)
29. New Orleans: -4.30 (-1.08)
30. Cleveland: -6.21 (-2.07)
31. Jacksonville: -6.63 (-1.66)
32. Chicago: -7.50 (-1.25)

I don't feel comfortable saying anything about those totals as a whole at this point for the aforementioned reasons.

I DO note that the league, as a whole, scored negative: a total of -51.28, or -0.53 per contract. A couple reasons why that might be:

1) The year-by-year adjustments might be off. Every year the salary cap increases, more money's available to spend on free agents; perhaps I haven't properly accounted for that change in the money available for free agents from season to season. Something to possibly futz with for next year.

2) The extra value might be made up by those sub-$3 million contracts. Just looking at it subjectively (as opposed to running the numbers), there were some good contracts signed for low-money contracts. Jelani Jankins at $1 mil, Bradley McDouglad at $1.8 mil, etc. I excluded those players because of potential false positives among "warm body" starters, but there's certainly SOME value there. Maybe a net positive!

3) The extra value might be made up by restructures/resignings not taken into account here. For example, the model loves the Tyrod Taylor restructure. It's a big fan of Eric Berry. If I remember correctly, it even thought the LeVeon Bell franchise tag was good value. Maybe the best deals this year happened before players hit free agency, and thus they didn't get onto that list. On a related note...

4) Maybe this year's free agency class was just bad! Or below average, at least. I didn't run this exact same model for previous seasons, so it's possible that next year, it will be at +0.5 per player, or something along those lines. Also, the three biggest free agent contracts went to a player past the age of 30, a quarterback with 11 pass attempts over the past two seasons and a cornerback AV hates because of a lack of starting experience. Could just be the model hating the particular set of players at the top of the model this year.

5) Maybe the problem's with the data, and not the model. It says each player is being over-paid by 0.53 AV, but the data that goes in is in --whole integers--. There's no such thing as fractional AV in the dataset! Why, if you round all the estimated values to the nearest whole integer, you get...uh... -51. And an average of -0.53.

OK, maybe that last one ~isn't~ it.

I don't have answers! Just theories.

10 Re: 2017 Free Agency Cost-Benefit Analysis

My 2 cents:
Your sample size is 97 players/contracts--it's not great, it's all you have--but that's less than the number of plays in a game. It's an average of 3 contracts per team. All it takes is a couple of outliers, which leads us to your 4th reason. That is part of the answer, combined with what you allude to in #3--there may be some "net-positive" contracts not in the data set.

Here's the statistical POV--players on positive-value contracts don't hit free agency--they get EXTENSIONS. So, there is realistically a "ceiling" on how high your "positive" FA contracts can go, but not the same "floor," because of overpays. Let's face it--A.J. Bouye wasn't going to get a 2 yr, $10M total deal--but that's what the model would rate as a good deal. So, it's almost impossible to have a deal that balances that "overpay"--unless it's less than $3M/yr--which your dataset excludes. So, while I understand your reason to exclude them, and it does have validity, it potentially eliminates some positive-value deals.

13 Re: 2017 Free Agency Cost-Benefit Analysis

Yeah, I think you're right -- it only takes one team to overpay someone on the free market, and very good players get locked up before that can happen, generally.

The sub-$3 million contracts get included in the dataset each season, so yeah, that's where a ton of the value likely is. I might lower my threshold to talk about some of those players next season -- or at least, highlight some particularly good deals.

Gotta automate data collection for that one, though; that's twice as many deals to talk about, and when I started doing this this year, we didn't have the 2016 AV in the database. So many PFR searches... so, so many.

35 Re: 2017 Free Agency Cost-Benefit Analysis

Looking over this very interesting piece (thanks for doing the work) made me think of something I've been chewing on for a couple weeks. I am a Chiefs fan and have absolutely 0% animosity towards the Miami Dolphins, but the contracts they handed out this off-season just struck me as insane. The #s on Alonzo, Stills, Sims,Woodley, Mitchell, giving their Safety a contract in Berry territory, etc. I know this is based on bringing in outside FAs, but any chance of cost-benefit analysis on team by team off-season spending? As many questionable deals that I thought were handed out by the "usual suspects" (Jacksonville/SF/CHI) I thought a team normally pretty well run(as much as I loathe them) Denver also had a mystifying free agency period, Carolina inked a couple baffling deals, Seattle is paying Luke Joeckal $8 million per, etc. But Miami with an improving youngish QB+A fantastic looking 2nd year HC seems to be playing with one arm tied behind their backs and it seems totally self inflicted.

8 Re: 2017 Free Agency Cost-Benefit Analysis

Anyone know what Malcolm Butler's AV is? If Butler doesn't resign with the Patriots, it will be interesting to see how his Age-Adj Val compares with that of Stephon Gilmore.

14 Re: 2017 Free Agency Cost-Benefit Analysis

Butler has had AVs of 1, 10 and 13 over the past three seasons, and he's only 27 years old. Perhaps unsurprisingly, the model loves him -- a projected AV of 10.22 for next season.

Only four defensive back in football have put up more AV over the past two seasons: Chris Harris, Marcus Peters, Richard Sherman and Aqib Talib. So, yeah, it basically says "give Butler all the money in the world"; it would predict a contract of $17.5 million a season if he was on the open market! That would make him the top-paid corner in the game.

That seems high to me by a rather large margin, but whether you think it should be $10 million or $12 or whatever, it's a heck of a lot larger than the $3.91 million tender he's getting as an RFA.

26 Re: 2017 Free Agency Cost-Benefit Analysis

From what I read, the MB situation and the Gilmore contract have a lot to do with height. Butler is a great cover corner but he's short, so he is great against OBJ or AB or guys like that, but not Julio Jones. BB wanted to add a taller CB1 even if it meant weaker coverage skills so that he could get better matchups. A 2017 CB depth chart of Butler, Gilmore, and Rowe would be really good with matchups against tall WRs, and that group would also be good against the run, which fits BB's Dline philosophy lately.

27 Re: 2017 Free Agency Cost-Benefit Analysis

Belichick let Butler go to the market because he figured the restricted franchise tag gives the Patriots a huge advantage. And that has turned out to be the case. No team wants to give Butler the money he thinks he deserves because doing so would be a double whammy: they'd both have a huge contract to deal with and they'd lose a 1st round pick to the Pats.

11 Re: 2017 Free Agency Cost-Benefit Analysis

Interesting, but it doesn't account for teams that re-signed their own FAs.
How do the Kenny Stills and Andre branch re-signings grade for example?

Phil Simms is a Cretin.

15 Re: 2017 Free Agency Cost-Benefit Analysis

It does account for them; they just weren't included in the article.

Stills is projected to get 6.38 AV, while his contract would be expect him to earn 6.56 AV. That's an AAV of -0.18.

Branch is projected to get 5.79 AV, while his contract would expect him to earn 7.28 AV. That's an AAV of -1.49.

18 Re: 2017 Free Agency Cost-Benefit Analysis

Thanks Bryan. Bill Barnwell in particular has torn Miami a new one over these two extensions from a value vs contract perspective, and it seems the AV analysis broadly agrees with him, especially on Branch. Still projects as basically a wash, but trading up to get Carroo and them having him as #4 WR doesn't seem a great use of draft assets...

Phil Simms is a Cretin.

21 Re: 2017 Free Agency Cost-Benefit Analysis

Yeah, if Stills was the lowest-scoring move Miami had put together, you could say "Oh, well, they see him in practice every day and have faith he'll improve, and decided that was worth an extra $400,000 a year". Defensible, I think, even if their wide receiver position is oddly crowded.

Branch, on the other hand...

16 Re: 2017 Free Agency Cost-Benefit Analysis

I think there is one factor that's missing from this analysis (and I don't mean that as a negative, because it would be very difficult to quantify), and that's the fact that teams must spend a certain amount per year whether or not there are players worthy of spending big money on.

This is the reason why despite the fact that I'm 99.9% sure that Glennon isn't a good QB, I'm not upset that the Bears signed him to the contract they did. $15M for a player of his caliber sounds like a ton, but there's a very low opportunity cost to the signing. They have to spend money on someone, and there are relatively few players in the league worth $15M a year, or even half that.

The key is that the contract basically commits them to only one year unless he plays well enough to stay on the team as the starter. If they had structured the deal so that they were committed to paying Glennon big money in future years when they might actually be a decent team, then it could be a bad contract. But for a bad team in the middle of rebuilding, it's almost irrelevant. (That's also why I think the Browns trade for Osweiler and a 2nd round pick was genius).

17 Re: 2017 Free Agency Cost-Benefit Analysis

It's definitely something to look into for future editions -- the existence of a salary ~floor~ as well as a salary cap means that money has to be spent somewhere, and teams with obscene amounts of room (Cleveland, Tampa Bay, San Francisco) can "get away" with worse deals simply because they have the excess room to eat it somewhat.

Then again, you'd also have to take into account how much salary can be rolled over to future seasons, and minimum cash spending per year, and when each team's target for being truly competitive actually is...this is the sort of thing that can get nearly infinitely complex if you let it.

I mean, Andrew, who wrote this article the last two years, now basically does this type of analysis for Cleveland directly. It's the kind of thing where you could get lost in the weeds in the specifics for each individual team, because you're right -- the value of a player for Cleveland is different than that for, say, Dallas, thanks to different salary obligations, different expectations, and even different philosophies. Infinite complications!

The best way to probably think about it is that no GM is out there ~intentionally~ signing bad deals. So, if a deal looks bad (like the Glennon deal does, on paper), it's just an opportunity to ask "why did they make this deal?" In Chicago's case, it's likely because they have roughly no expectations going into this season, aren't sold on any of the quarterbacks at the top of the draft, had excess salary room to spare, have a deal they can slip out of after a year if things go terribly wrong, and feel like they have a better chance to get a $15-million player out of Glennon than they would out of other players in general. Now, I think they're ~wrong~ on that last bit, but you can at least see the strategy there.

...Still, I'm fairly sure I'd rather have Hoyer for $6 than Glennon for $15. But then, that's why I don't get paid the big bucks.

22 Re: 2017 Free Agency Cost-Benefit Analysis

Yeah, I definitely didn't mean to suggest that signing players to contracts that aren't wildly out of line with their skills goes out the window with a Cleveland or a Chicago. I would like to think that Ryan Pace and company think Glennon stands a decent chance of being worth $15 million/year; whether they're right or wrong in that assessment is a different question.

I agree with you about Hoyer vs. Glennon if it's strictly a value comparison based on how they're played so far. I do think that Glennon has some nonzero chance of being their long-term answer at QB that Hoyer does not. I'm not sure that chance is worth $9M, but given that they can't hoard the extra cap space and use it in the future when they've (hopefully) got a bunch of great players coming off of rookie contracts they need to resign, I see the argument that it doesn't have to be much of a chance at all to take the gamble.

Actually one of the most mystifying things about Glennon's contract to me is that it sounds like the Bears were kind of negotiating against themselves. I'm wondering if maybe $12-13M a year would have been enough to sign him, but maybe there was another team out there that the Bears had to outbid.

24 Re: 2017 Free Agency Cost-Benefit Analysis

I can confirm, right here and right now, that I would be a ~terrible~ choice to help run an NFL franchise. I mean, I'd do it; I'm right here and willing to get to work. But the decision making process that would lead to a franchise hiring me would be indicative of bad decision making processes across the board.

25 Re: 2017 Free Agency Cost-Benefit Analysis

I'd be curious to see how poor you would be? I suspect a good chunk of the job is really politics. Lots of it.

Outside of a few teams, most nfl front offices feel like they are run the same way. You hire a coordinator from a successful team as the head coach. The coordinators are usually filled out with either ex head coaches or high quality college coordinators. Then the rest of the staff are either holdovers from the past or purgers from the head coaches last spot.

Drafting surprisingly follows most of Mel Kiper's board, especially in the first round. Depending on your level of urgency/relationship with the owner - you make a franchise changing trade where you mortgage the future for a specific player or you sit tight, draft the safe boring choice, and go forward.

Would you do something different? Would you be willing, say, to trade your 3s and 4ths for more 5ths and 6ths because draft value models show almost no difference in quality between 5th and 6th vs 3rd and 4ths(controlling for position).

Would you eschew drafting a running back altogether until round 3. Ditto for Kickers, special teamers, full backs, etc etc. Would you even be willing to pass on a left tackle if you had a top 5 pick? Gutsy, but there's been no evidence to show that left tackles juice up an offense for over a long period of time. In fact, the game seems to be showing o line quality is a lot more about depth than one great anchor.

28 Re: 2017 Free Agency Cost-Benefit Analysis

The Burkhead situation is interesting. His largish contract suggests to Pat fans that he'll be seeing a lot more snaps in New England than he did with the Bengals, where he was buried on the depth chart. Right now it looks like he'll be getting the snaps from Blount, who remains unsigned. If that's the case, then projecting his AV from his prior seasons would make little sense.

But, of course, maybe he won't.

29 Re: 2017 Free Agency Cost-Benefit Analysis

Yeah, in a ~very small sample size~, Burkhead was very effective last season. AV gives him some credit for that, but also notes that he's been a backup, and doesn't give him any extra credit - after all, if he was really any good, he'd be starting, right?

Well, maybe not. When dealing with a player going from a backup role to a potential starting one, scouting is a far more useful tool.

Burkhead's contract assumes he'll be a 5 AV player, essentially. Here are the list of 5 AV players from last season:


So, another way of phrasing it is that the Patriots are gambling that Burkhead will get them about 800 yards of offense -- or, to put it another way, that he'll do this year about what James White did last year. Obviously, he's never done that in his career, but it's not subjectively a crazy gamble.