What Are Teams Paying Per WAR in Free Agency?
After a quiet 2020 offseason, and in advance of the ongoing lockout, the early 2021-22 free agency period saw a sudden burst of activity. Teams shelled out more than $1.5 billion in new contracts, a record-breaking pace. Not only did they act earlier in the winter than we’re used to, they also spent far more than last offseason. Is free agency fixed? We’ll need to dive into the data to find out.
See, “how much money was spent on free agents” is an inexact measure of teams’ spending appetites. Imagine an offseason where, due to strategic contract extensions and a wildly immoral use of cloning technology, the only players on the free agent market are 37 versions of Alcides Escobar and 25 copies of Jordan Lyles. Free agency spending would crater, and it would be hard to blame teams for it. It’s not as though you have to give the best player on the market a $300 million deal; contracts are, obviously enough, affected by the caliber of player signing the contract.
Rather than come up with some new form of analysis, I decided to use a methodology advanced by former FanGraphs writer Craig Edwards. The idea is straightforward: take players projected for 2 or more WAR by Steamer in the upcoming season, apply a naive adjustment for aging, and project how much WAR each free agent will accrue over the life of their contract. Like Craig, I applied some discounting for playing time projections. That lets us create expected $/WAR numbers for each year’s free agency class:
| Offseason | 2+Proj WAR |
|---|---|
| 2018 | $9.3 M/WAR |
| 2019 | $7.8 M/WAR |
| 2020 | $9.5 M/WAR |
| 2021 | $5.5 M/WAR |
| 2022 | $8.5 M/WAR |
Last year’s free agency class had it rough. With teams licking their wounds after an abbreviated and crowd-less season, they offered far less in free agency. The raw numbers show how little money was being thrown around, and even that understates it. Out of the 18 free agents projected for 2 or more WAR, only five received contracts of three years or longer. Not only were teams hesitant to spend on a per-win basis, they stayed away from long-term deals while doing so.
The early results from the current free agency period suggest that this decline was a one-off. Before the 2020 season, 12 of the 15 contracts signed by players who were projected for 2 or more WAR were for three years or longer. This year, 14 of the 20 similarly-projected free agents who have signed received three years or more. After a brief blip of short-contracts-only thinking, the length of contracts appears to be rebounding.
Despite that rebound, teams are still paying less per win on the open market than they did before the COVID-shortened 2020 season. The total outlay this year might be higher than, say, the Gerrit Cole/Anthony Rendon/Stephen Strasburg bonanza that took place in the pre-2020 offseason, but that’s because the players who have signed project to accrue more WAR. Merely among players who have already signed – which excludes 10 of our top 25 free agents – the players projected for 2 or more WAR in 2022 project to be worth roughly 175 WAR across the length of their contracts. That number checked in at 135 in 2020 after all was said and done.
In other words, this year’s class is deservedly getting paid for its future contributions. That helps explain both why so many teams have needs in free agency and why so many deals have been signed. Teams lost good players – better than the ones who reached free agency in any of the last four years – and there were good players available in free agency. Absolute dollars spent only matters so much, and this year’s overall outlay is mostly in keeping with previous years after accounting for that.
While we’re here, we might as well update Edwards’ other investigation: is the cost of WAR still linear? I used the same methodology as his last study. The key step here is removing all reliever contracts, as teams compensate them differently (read: more money per projected WAR) than starters and position players. Since there are rarely (if ever) any relievers in the 2+ WAR cohort, removing them makes the comparison more even. Here’s that analysis:
| Offseason | 2+Proj WAR | 0-2 Proj WAR, No RP | Difference | % Difference |
|---|---|---|---|---|
| 2018 | $9.3 M/WAR | $8.0 M/WAR | $1.3 M/WAR | -13.8% |
| 2019 | $7.8 M/WAR | $5.3 M/WAR | $2.5 M/WAR | -32.1% |
| 2020 | $9.5 M/WAR | $8.0 M/WAR | $1.5 M/WAR | -15.8% |
| 2021 | $5.5 M/WAR | $4.8 M/WAR | $0.7 M/WAR | -12.7% |
| 2022 | $8.5 M/WAR | $6.5 M/WAR | $2.0 M/WAR | -23.5% |
Of note, my data for 2020 low-WAR players differs from Edwards’ previous finding. He had that group at $6.7 million per projected WAR. My guess is that this is due to the number of low-WAR players who had not yet signed when he ran his analysis. Several of them projected for 0-ish WAR; adding those players increased the dollar-per-WAR metric there. It’s also possible that our cohorts and WAR projections differed; I used Steamer exclusively, whereas Craig may have used a Depth Charts blend. In addition, the numbers for this offseason could still change meaningfully — the signed 0-2 WAR group only numbers 19 players so far, while it frequently ends up numbering in the low 100s in regular years.
What does this mean going forward? There still appears to be a meaningful discount applied to role players and part-time contributors. While salaries are rebounding, they haven’t even reached the levels that prevailed before the pandemic, much less outstripped them. Teams might yet shell out record contracts this offseason. As of now, though, they haven’t – and they’ve continued to pay a premium for stars while the middle class of free agency gets a discounted rate.
Ben is a writer at FanGraphs. He can be found on Bluesky @benclemens.
“What does this mean going forward? There still appears to be a meaningful discount applied to role players and part-time contributors.”
Maybe it’s semantics, but I think this is more about a premium being paid for concentration of WAR in a single player / roster spot. A four-win player is more valuable than two two-win players because of the potential for additional contribution at the additional roster spot (as well as the scarcity of roster spots).
It speaks more to the issue of using replacement player as benchmark to value. If it were “wins above average” or “wins above replacement starter” we probably have more consistency
I doubt it as the below average players making money would mess it up. WAR seems linear to me for most ranges except really close to zero and really far from zero. Around zero, the differences in projected WAR and the estimate of replacement level drive up the variance. For far from zero, teams pay a little surplus for the elite players and some perceived elite players that would still not be linear for wins above average.
WAA = WAR – 2.0 more or less.
It moves the zero axis, and then we say that teams pay $/WAR for above average players at a higher rate than they do for below average players. Which…is the same thing
I’m going to respond again, because I’m pedantic 🙂
First, a player who accumulates WAR <0 has pretty clearly had a bad season. Theoretically a team could have improved on that player by simply plumbing the depths of their AAA squad or making a minor trade for another org's surplus. This player actually hurt his team's chances of winning ballgames, and that's unquestionably bad.
Now, a player who accumulates between 0-2 WAR did not necessarily have a bad season. He most certainly posted below average aggregate results, but he did help his team win. Albeit less than the average MLBer did. Maybe he was a mediocre player who chipped in minuscule bits over the course of a long season, but lots of 0-2 WAR guys are good players with small roles, or good players who missed time due to injury. For one example, MLB is overrun with terrific bullpen arms who garner less than 2 WAR because they don't throw enough innings to accumulate more than that, but absolutely help their teams to win. For another, a team may have a 4th OF or 2nd C who would start for most of the rest of MLB. He performs at a high level when he does play, but only gets 200 PA on the season because he's buried on the depth chart. Those guys all earn positive WAR, but maybe not enough to translate as positive WAA.
When you see a guy with negative WAR, it's easy to shorthand it as "this guy sucked". The tendency with negative WAA would be the same, but there's a lot of room just under 0 WAA to have guys who actually performed at a higher than average level but over a lesser span of innings.
On the team level, however, WAA might be more useful. A 0 WAR squad would, it's supposed, win about 45 games in a season. That's a weird and arbitrary place to be. A team that was 0 WAA, however, would post something close to a .500 record.
The Giants in 2021 seem to have screwed all that up. Look at pre and post season WAR accumulation.
That’s the general assertion, but how true is it? It’s extremely difficult to find eight productive, 2 WAR or better players for your lineup, and this also creates the problem of paying the additional gentleman you’ve made room for.
In addition, your 4 WAR player is creating room for… who, exactly? This will probably vary team to team. For example, the Tiggers might sign a 4 WAR 2Bman on the grounds that this lets them try a rookie in the OF, gambling on his production, and minding less if he puts up replacement level production since you’re able to project getting 4 WAR instead of 2 WAR by signing, let’s call him… Javy.
Still, you’ve created other problems for yourself. Instead of betting something like two 2/20m-ish contracts for, I dunno, let’s call them Eduardo and Mark, you’re committing to a couple of expensive years at the end of Javy’s 6/140m deal you really, really don’t want.
In short, I’d be curious whether there’s any compelling evidence that getting 4 WAR from one roster spot is actually better, overall, when all factors are included, than getting 2 WAR each from two roster spots. We’d probably have to start by seeing what the historical price per win turned out to be for projectable 4 WAR FAs.
Two things that may be worth considering…
-There are approximately 5x as many players projected for 2 wins in Steamer (35 or so) than projected for 4 wins (7 or so). So if there is a small advantage then scarcity will magnify this difference.
-A 2 win player doesn’t actually improve several rosters, taking them out of the running entirely for 2-win players. And those rosters also tend to be right at the cusp of needing a talent infusion to compete for the playoffs–say, taking an 85 win team to an 88 or 89 win team. So there’s a lot of motivation for those teams to snap up 4 win players and zero for them to go after 2-win players.
That said, there are definitely “stars and scrubs” rosters where two 2-win players should be as advantageous as one 4 win player. The Guardians and Rangers are the two foremost examples. It’s very nice that they have a star position player or two now, but at least at the moment it doesn’t seem more valuable than spreading it around to different positions.
This would matter more if bad teams actually cared about finishing 81-81. But they would rather tear down and aim for a future 90-95 win team. Their fans prefer this too. So the real market for 2 Win players are the Win Now teams with holes and money to spend. That’s way less spots than the total MLB starters.