2018 Free Agency Cost-Benefit Analysis

by Bryan Knowles

You can be forgiven for feeling that free agency was a bit ... underwhelming this year. There was plenty of cap space to go around, but not a ton of crème de la crème players to use it on. Following a nigh-unprecedented number of high-level trades and preceded by a draft season filled with interesting debates over the value of running backs and the stats-versus-"eye test" conflict, free agency didn't make as much as a splash as we normally expect.

That doesn't mean there's nothing to talk about, however. Kirk Cousins has boldly gone where no man has gone before, receiving the largest contract in NFL history and having it all guaranteed, to boot. The Jets added nearly $200 million to their books, signing 18 free agents including Trumaine Johnson. The Bears added more than $100 million in guaranteed money. All that extra cap space had to go somewhere, so it's time for our annual attempt to identify the best and worst values of the free-agent market.

Our methodology is mostly the same that we used last year. Using Pro Football Reference's Approximate Value statistic as our measure of player quality, we perform a regression analysis that looks at each player's previous three years of performance and their age. This allows us to create a rough approximation of their future performance, based on how similar players either developed or declined over the course of their careers. The word "approximation" is doing a lot of work here, but in terms of comparing players across years and positions, it's a very useful tool.

We then compare that projected value with the implied value of the new contracts -- that is, what AV a player would have to put up to justify their contract, based on how teams have divvied up their salary cap dollars in the past. That allows us to calculate each contract's Age-Adjusted Value, our estimate of the value that each contract creates or destroys.

You'll also note that there are more net-negative deals listed here than positive deals -- by about a 2:1 ratio, at that. We're not saying that free agency is a net-negative proposition, nor are we saying that a negative-value deal is necessarily bad. There are a couple reasons why the tables below have more negative players.

Firstly, the tables below include every free agent with a deal worth more than $3 million per year. That cut-off, however, eliminates more than 150 free agents who signed smaller, usually shorter deals. Teams can get a lot of value out of those cheap signings. Players like Case Keenum, LeGarrette Blount and Zach Brown provided significant value for their teams last year after signing shorter, smaller contracts.

Almost by definition, the more you pay, the less likely you are to find excess value. Rather than have massive tables filled with players making the veteran's minimum, we selected a somewhat arbitrary cut-off for "significant" signings for the main portion of the article, and we'll highlight some of the best cheaper deals in a supplemental table at the end.

Secondly, not all overpays are necessarily bad. Take the Kirk Cousins contract for example. At $28 million per year -- all of it guaranteed -- it's the largest contract in the NFL. It's a deal you'd expect someone like Russell Wilson or Matt Ryan to receive, and Cousins hasn't been on that level. But Russell Wilson and Matt Ryan were not available in free agency. Nor were there multiple Kirk Cousins-type players to bid on to drive the overall prices down. Minnesota had to overpay to get the best quarterback available. Essentially, they traded financial strength for positional strength -- they have less money to go after other players, but no longer have to worry about the quarterback position.

You can't build an entire roster like that -- you'll end up with no depth, having spent all your money on big-name free agents. But splurging on a few big-name players can be beneficial if you have found savings elsewhere. This is why teams like the Rams, Eagles, and Cowboys have a significant advantage in free agency -- they're essentially saving millions by having a high-quality quarterback on a rookie contract, allowing them to be a little less thrifty in the free-agent market during that window of opportunity. They can overpay for the likes of Ndamukong Suh and take the hit more than the Vikings, 49ers, or Lions could. That opportunity can be wasted if, say, you have $22.8 million in dead money, but the point is that a deal is only good or bad when considered in the context of the team's overall salary structure.

Each table will have a player's new deal details -- their total value, value per season, and amount of guaranteed money. We list both their projected AV -- the amount of value our regression indicates they'll produce going forward -- and their needed AV -- the amount of value their contract implies. Then, you'll have the Age-Adjusted Value, which is how the model grades each move. Finally, you'll have words which explain where the model might be overly optimistic or pessimistic on specific players.

Quarterbacks

Quarterbacks are a little tricky to predict, because their value is exceptionally binary: you either are starting and producing value, or you're on the bench and not. There's no rotational quarterback who puts up a little bit of value and gradually works his way into a starting role. Projecting someone going from a backup job to a starting job is one of the hardest things a general manager has to do, and quarterbacks have some of the smallest sample sizes to work with.

At an expected 16.4 AV per year over three years, Kirk Cousins would need to have 49 AV over the next three seasons to match his new record contract. Over the past three years, the quarterbacks with the most AV have been Tom Brady, Matt Ryan, and Russell Wilson -- all of whom have put up 49 AV. The best quarterback in the league should be the highest paid player in the league. Cousins is obviously a few steps behind those guys.

But those guys weren't available, and Cousins' 13.38 AV-per-year projection makes him the third-best quarterback free agent under the current CBA, stretching back to 2011. He's only topped by 2011 Peyton Manning and this year's Drew Brees, both of whom were always going to re-sign with their current teams. Cousins is simply the best quarterback to have been openly available in recent memory, so it's not surprising that teams backed the Brinks truck up to his front door.

The quarterback list also has a few examples showing why a team may have been willing to overpay the straight projection. First, some teams are expecting a recent improvement to be sustainable, and not a one-year fluke. If the Broncos have signed 2017 Case Keenum (AV 14), then they're paying a little under market value. His projection is hurt because he wasn't a particularly good quarterback with the Rams under Jeff Fisher, but if we learned anything last season, it's that quarterbacks improve when they're not being coached by Jeff Fisher.

Second, some teams believe that someone is going to recover from a serious injury. If Teddy Bridgewater returns to his pre-injury form, the Jets have the steal of the free-agency class. Obviously, that's a huge "if" there, and any injury that costs you a season and a half is extremely serious, but Healthy Teddy Bridgewater was worth somewhere in the $15 million to $20 million range.

Third, some players might outperform normal aging curves. Drew Brees' deal is only negative because he turned 39 in January; age catches even the best players eventually. There have only been two quarterbacks 39 or older who have put up 15 AV in a season -- 2009 Brett Favre and 2017 Tom Brady. That being said, Brees has passed 15 AV in each of the last two seasons, and the Saints have a better feel for Brees' specific situation than a projection does, so there's no reason to believe Brees' deal can't pay off.

On the other hand, Chase Daniel throws three passes in three years, and Chicago hands him $5 million guaranteed at signing? Man, how do I get that job? This is the second year in a row the Bears gave out the worst quarterback contract, after signing Mike Glennon last season. You can get perfectly serviceable backups for much less than Chicago will be paying Daniel.

In Arizona, Sam Bradford's deal would be at about a Kirk Cousins level of overpay if you could trust him to stay healthy. The model does not trust him to stay healthy.

Running Backs

A good year to sign Patriots running backs. The model doesn't know that Rex Burkhead will likely be getting a larger share of the snaps with Dion Lewis out of town, so if anything, it's undervaluing his potential going forward. I think I'd still prefer Lewis' deal over Burkhead's if given the choice -- Lewis is the better player, after all -- but neither the Patriots nor the Titans come out looking bad here.

Jonathan Stewart has finished with 5 AV in each of the past four seasons, but he turned 31 in March -- he's at the age when running backs turn back into pumpkins, hence the negative projection. When healthy, Carlos Hyde is worth what the Browns paid to get him, but he has only played 16 games in one of his four NFL seasons, so that's a worry as well.

But there's only one real questionable deal here. For the second year in a row, the 49ers have come out worst in our model -- last year for Kyle Juszczyk, and this year for Jerick McKinnon. McKinnon is now the fourth-highest paid running back in football, despite never gaining 1,000 yards from scrimmage in a season. He's a better fit for Kyle Shanahan's offense than Hyde was, but this seems like paying for a projection rather than what McKinnon has shown he can do on the field. McKinnon's contract is very front-loaded, so the 49ers can get out of it after one year if things go south, but it's a questionable year nonetheless. McKinnon did hit 9 AV last season, so it's not out of the question he can pass the 11 AV his contract implies. Still -- given a deep running back draft class available next weekend, did the 49ers have to splurge so much on McKinnon?

Wide Receivers

Yes, the model is panicking over Allen Robinson's deal because he essentially missed all of 2017 with a torn ACL. The model does account for injuries -- it doesn't assume a player is suddenly valueless just because they got hurt -- but it currently does not distinguish between an injury like Robinson's, which cost him a year but shouldn't impact him going forward, and the injuries suffered by Teddy Bridgewater or Zach Miller, which cost multiple years or could be career ending. That's something we'll look at improving in next year's iteration of the model. If you assume Robinson's 2017 season would have been as good as his 2015 breakout season, the model would project him between 9 and 10 AV -- a moderate overpay for the best receiver in the class.

We don't have an injury defense to explain the deal for Sammy Watkins, who is now the fourth-highest paid receiver in football. The Chiefs are paying him to be Julio Jones or Antonio Brown, and Watkins has never really come close. He battled injuries in Buffalo and wasn't a good fit in Sean McVay's scheme in Los Angeles, but is that enough to justify paying him as a top-five receiver? I'm just not sure.

One thing I am sure of, however, is that the Mike Wallace deal looks great for Philadelphia. They've basically managed to upgrade from Torrey Smith while paying less for the privilege. Yes, Wallace's age is a concern, but it's a one-year deal and the Eagles are in win-now mode. Wallace should be a nice fit as Philadelphia's deep threat.

Everyone from Allen Hurns through to Taylor Gabriel can look like a good deal if you judge a player by his highs rather than his lows -- every receiver in the group had at least one year in the last three where they surpassed their projection. My favorite of the group is Danny Amendola's deal in Miami -- a cheaper upgrade over Jarvis Landry, albeit one with significant injury concerns.

Oh, and though they're not on the table: the model would project Dez Bryant to be worth a little less than $8 million a season going forward, and thinks Jarvis Landry should have been worth about the same to Cleveland. (The Browns traded a fourth-round pick this year and a 2019 seventh-rounder to acquire Landry, then signed him to a five-year extension worth up to $75.5 million, $15.1 million per season, with $34 million guaranteed.)

Tight Ends

You'll note that the AV expected of tight ends is somewhat lower than that of the other skill position players. This is a problem with AV -- it tends to underrate those tight ends who block more than they catch. That lower rating gets baked into the regression, so tight ends come out with lower numbers than you'd expect to see elsewhere.

Eric Ebron drops too many passes and needs to find more consistency in all aspects of the game. That being said, he's only 25 and has shown he can be a solid underneath receiver who makes plays after the catch. He'll frustrate Colts fans like he did Lions fans in Detroit, but he has also proven he can be a valuable player, just not the first-round value the Lions thought they were getting when they drafted him in 2014.

When healthy, Tyler Eifert was a Pro Bowler. He hasn't really been healthy in two years, but if he can return to anything approaching form, than Cincinnati has a great deal here.

Jimmy Graham is an interesting case; he never seemed to really click in Seattle, and his yards per reception plummeted to 9.1 last season. If Graham returns to his New Orleans form with the Packers, then this is a solid deal. Green Bay is hoping that Graham's struggles were with the scheme rather than the player, in other words.

The model hates the Trey Burton deal because he has only started five games in his career; the Bears will presumably be using him just a bit more if they're paying him $8 million a year. Based on his DVOA from last season (35.0%, third-best at the position), the deal is far more justifiable.

Offensive Linemen

At this point, Breno Giacomini is a consistent presence you can slot in at right tackle for adequate performance. A one-year deal for $3 million for Giacomini isn't going to create major splashes, but it's a cheap move that's relatively unlikely to backfire. If I were rating these subjectively, I'd prefer Brandon Fusco's deal in Atlanta, but to each their own.

Andrew Norwell was the best offensive lineman available, and the third-most valuable player, period, behind Cousins and Brees. The projection system is a little hesitant about him, because he jumped from a 6 AV player in both 2015 and 2016 to a 14 AV player last season. Norwell is now the highest-paid guard in the league, and the model is asking him to play like the best guard in the league to justify it. Norwell's 14 AV was the highest in the league last season, so if he can keep up that level of play, he'll be worth every penny.

Nate Solder is now the highest-paid tackle in football by nearly $2 million a season. That's a lot of money for any tackle to justify. I do, however, think the model is expecting too much out of a $15 million offensive lineman, simply because it hasn't had any examples close to compare in the past. It does take more money to jump from 10 AV to 11 AV than it does from 5 AV to 6 AV, but I think it might be accelerating too sharply here to try to model an unprecedentedly high lineman contract. I'd pick Weston Richburg as the worst deal, simply because he's coming off of a very serious concussion.

Defensive Linemen

We're mixing interior defenders, 3-4 defensive ends, and 4-3 edge rushers here, because that's how the data comes to us. One day, the NFL will make edge rushers a separate category for franchise tag purposes and the like, but until then, we shall muddle through somehow.

Edge rushers are generally paid more than interior defenders. They also generally earn more AV, so some of that ends up being a wash, but 4-3 ends should be a little higher on this table than they actually are.

Julius Peppers is only the worst deal listed here because, statistically speaking, he is dead. He turned 38 in January, and there have only been about a dozen defensive linemen in NFL history who have been solid contributors at that age. Pat Williams, Ted Washington, and Bruce Smith are the only three examples this millennium. The model sees a 38-year-old defensive lineman and assumes he should be fitted for a walker and a hearing aid, not penciled in for double-digit sacks. Age will catch up to Peppers eventually, but the model has penciled him in way too low. There's no way he's the worst deal on this list.

Nor is Ndamukong Suh the worst deal -- I think that's an outright problem with AV. Suh only earned 7 AV in both 2015 and 2017, because there wasn't a lot of value to spread around on some poor Miami defenses. In both those years, no one on Miami's defense earned more than 9 AV, and Suh was in the top three each time -- what little value Miami was generating was going to Suh. I think it's fair to argue that Miami Suh hasn't been as good as Detroit Suh, and age is a concerning factor for the long run, but I strongly feel AV is underrating Suh's value. I suppose that leaves Chris Smith as the largest overpay, but it's not an egregious one.

Instead, there are a lot of good value deals here that the model really likes. Bennie Logan is somewhat inconsistent on a snap-to-snap basis, but his ball awareness, pursuit and tackling ability had him among the league leaders in tackles behind the line of scrimmage last season. Chris Baker appears near the top of the list for the second year in a row, and I'd highlight Muhammad Wilkerson, Sylvester Williams, and Sheldon Richardson as particularly good values, as well. I also think AV is underrating Haloti Ngata; at age 34, his best days are behind him, but a $3 million deal seems like good value for an experienced veteran.

Linebackers

Again, we're mixing 3-4 rush linebackers with inside linebackers here, so the same caveats apply.

Christian Jones has a high projected AV because someone on the Bears had to. A defensive player's rating will be largely based on how many games he played, how many games he started, and how good his team was defensively. Over time, the theory goes, better players will play more games on better defenses, and thus get a higher rating. That seems fairly intuitive when looking at a large enough sample. When looking at single-year stats, however, sometimes a player gets snaps not because he is particularly good, but because there's just no one better to take those snaps.

Enter Christian Jones. Jones is a backup who was forced into a starting role in both 2015 and 2017 thanks to issues in the Bears' linebacker corps, like pectoral injuries to Jerrell Freeman and Nick Kwiatkoski in 2017. AV sees Jones as a starter, who should get starter value. You should probably view him as a backup and downgrade him accordingly. That would bump Preston Brown up to the top slot -- $4 million for a consistent, technically sound, jack-of-all-trades inside linebacker is a good deal.

At the bottom of the table, Trent Murphy missed all of 2017 with a torn ACL and MCL. Even giving him the most credit possible and assuming he'll come back in full form with no lingering effects, it's hard to see $7.5 million of value here. He lacks the explosive pass-rush traits you'd look for out of your top pass-rusher, though he did pick up nine sacks in 2016.

Cornerbacks

The highest projected AV belongs to Malcolm Butler, and he received an appropriately large contract to match. The deal for the Titans will look even better if Butler performs at the same levels he did in 2015 and 2016. It's rare that the top player at a position doesn't get overpaid, but the Titans have managed that with Butler. Butler received $11.25 million less overall and $10 million less guaranteed than the other big name free-agent corner, Trumaine Johnson. While I think Johnson's projection is being somewhat lowered by the Jeff Fisher effect, there's certainly not a $2 million a year difference between the two players.

The argument in defense of the Aaron Colvin signing is that it's only $18 million in guarantees, but even that feels too high. Colvin benefited from being surrounded by excellent athletes in Jacksonville. He's not good in man coverage and is not good at playing the ball. He's a zone defender and a good run stopper, but it feels like he got over-paid for his association with a top defense last season.

Both Ross Cockrell and Orlando Scandrick have out-performed their projected AV in each of the last three seasons, making their new, sub-$3.5 million deals feel like real bargains. Cockrell led all qualified cornerbacks with a 77 percent success rate last season. Neither player will be confused for a top shutdown corner, but as complementary pieces, they offer great value at a very reasonable price.

Safeties

There aren't any substantially notable deals among the safeties this year. Tyrann Mathieu may grade out as the worst deal, but it's not a significant overpay for the top safety in the free agent pool.

Special Teams

It's worth giving up a significant portion of your salary cap space to lock up, say, Justin Tucker. Matt Bryant, Greg Zuerlein, Adam Vinatieri -- these kickers have shown significant value above average over multiple seasons. Chandler Catanzaro and Cody Parkey do not fall in that group.

Smaller Deals

As promised, here are some of the best values found for less than $3 million. Specifically, these are the most valuable signings at each position group for players making more than the veteran's minimum:

Tom Savage has not looked good on the field, but $1.5 million for a quarterback with starting experience? In a world where Chase Daniel makes $5 million? That's a bargain. Deonte Thompson has speed, and caught 27 passes for 430 yards and a score for Buffalo in just 11 games. Kony Ealy has been inconsistent in his career, but had 23.5 pass pressures last season; he can be a useful part of a rotation. David Amerson wasn't worth the big deal he received from Oakland in 2016, but a one-year "prove it" deal as a starting corner in Kansas City to see if he can regain any of his exceptional 2015 form is more than worth it. If anyone can come up with a logical explanation as to why Marquette King was cut by the Raiders, I'd love to hear it.

All AV figures courtesy of Pro Football Reference. All salary information courtesy of Spotrac.