What Are Groups Paying For A Win In Free Company? 2026 Version

What are groups paying for a win in free company? Earlier this month, I answered a FanGraphs Weekly Mailbag query about that very subject, outlining a rule I’ve been utilizing in formulating my contract predictions. I left my rationalization unfastened and obscure as a result of it was certainly one of 4 questions in a mailbag, however to provide the normal gist, I take into consideration free agent salaries on a graduated scale, with function gamers being paid much less per win above substitute than superstars. Right this moment, I’d prefer to again up my argument with a bit extra mathematical rigor.

One of many advantages of writing for FanGraphs is that good baseball thinkers learn the location. I awoke final Monday to a direct message from Tom Tango, MLB’s chief knowledge architect. Tango had a number of strategies for additional analysis, a way for adjusting previous years of information for present payroll conditions, and even a hyperlink to a dialogue of the price of a win with Sean Smith. Smith, higher often known as Rally Monkey, is the creator of Baseball Reference’s calculation of WAR – once you see rWAR, that really stands for Rally WAR, not Reference WAR. In different phrases, I received assist from some heavy hitters.

With Smith’s glorious article on free company as a information, I constructed my very own methodology for analyzing the offers that free brokers obtain and turning them right into a mathematical rule. I took each beginning pitcher and place participant (relievers are bizarre and must be modeled in another way because of leverage issues) and famous their projected WAR within the subsequent season, in addition to the size and phrases of their contract. I excluded gamers who signed minor league offers, had been projected for detrimental WAR, or whose contract particulars had been undisclosed. To offer you a way, making use of this strategy to the 2025-26 offseason leaves us with 89 gamers, from Kyle Tucker all the way in which all the way down to Jorge Mateo.

I then used a components, calmly modified from the one Smith makes use of, to deal with future years. I assumed a decline of 0.4 WAR per yr on all projections because of ageing. Smith used a 0.5 WAR decline, however 0.4 is the imply decline throughout the multi-year ZiPS projections I used most lately, so I went with that. This tends to depress the modeled $/WAR value of stars, who signal longer contracts with a much less precipitous WAR decline, however I believe it displays actuality higher. I additionally utilized a 5% annual inflation charge to future yr ensures. That’s the speed Smith used, and it is sensible to me. I put all the pieces by way of 2025 {dollars}, too, simply to place every year on the identical scale. For the report, this inflation-and-aging technique is what I take advantage of in my very own contract modeling and in participant analysis for our annual Commerce Worth Sequence – I simply attempt to get higher measures of ageing straight from Dan Szymborski. I believe it’s a terrific theoretical basis.

With this knowledge in hand, I went about analyzing it in two alternative ways. First, I simply added up all of the projected WAR on one hand and all of the inflation-adjusted assured cash on the opposite, and divided the second by the primary. That’s so simple as it will get, with one dollar-per-WAR stage for everybody. Nicely, nearly so simple as it will get: I subtracted the minimal wage for every year, for causes I’ll clarify later.

The second technique makes me extremely grateful that I learn Smith’s piece on the price of a win. Should you’re conversant in his work, you recognize that he has a knack for chopping to the core of the difficulty and lopping off unnecessary complexity. That’s what occurred right here. I’ve a sophisticated contract mannequin that I take advantage of to make player-by-player predictions yearly, and I approached the query of how a lot a win prices in free company from that angle. However there’s a neater approach to do it, and I beloved Smith’s framing a lot that I used it to information my evaluation.

I divided gamers into tiers based mostly on their Steamer-projected WAR within the first yr of their new offers. I calculated the identical dollar-per-WAR quantity from up above individually for each tier, with the identical minimal wage adjustment from up above. I went with tiers of 0-1, 1-2, and 2-plus WAR for the sake of simplicity, although I additionally examined different breakpoints with related outcomes, together with a check the place I excluded the highest free agent from every year to keep away from outlier issues. It’s simpler than understanding completely different prices for various marginal wins, and the outcomes had been telling in each case.

Clearly, every of those strategies is aware of how a lot complete cash was spent in free company, so that they each have the full spending stage completely appropriate. On condition that I used precise contracts as a information, it couldn’t be every other manner. They carry out in another way in describing particular person contracts, nevertheless, and the three-tier mannequin does significantly better. To clarify the distinction, I’ll first current the outcomes from the 2025-26 offseason, then comply with with an expanded technique that I used to deal with a pattern of the final seven offseasons, way back to the RosterResource database goes.

Within the 2025-26 offseason, 19 gamers projected for two or extra WAR by Steamer signed contracts in free company. They obtained an combination $1.866 billion in ensures. After accounting for the size of their contracts, utilizing Smith’s inflation-and-aging technique of valuing future years, they obtained $12.84 million per projected WAR. Steamer tasks one other 29 of this winter’s free brokers for between 1 and a pair of WAR. They received paid $8.51 million per projected WAR. Lastly, 41 free brokers challenge for lower than 1 WAR in 2026; they obtained $6.74 million per WAR. That’s the three-tier mannequin. The one-tier mannequin would let you know that there have been 89 free brokers, and that they obtained $11.23 million per projected WAR.

Now, each of these issues are true. I don’t discover them to be equally descriptive, nevertheless, and the numbers again me up. For one factor, the three-tier technique has a imply absolute error of $8.5 million, as in comparison with $9.7 million for the straightforward rule. However to get to extra measures of statistical curiosity, we’ll must broaden out the pattern.

I took each free agent contract within the RosterResource database going again to the 2019-20 offseason. For 2020, I pro-rated the projections as much as a full season; the contracts gamers signed in that yr had been signed earlier than the COVID interruption, and I’d use the full-season projections even when they weren’t, as a result of a 2020 WAR projection for a 60-game season doesn’t play good with ageing curves except you exchange it right into a full-season projection. I took the projected pitching and batting WAR for all gamers and added them as much as account for two-way gamers – thanks, Ohtani, for the additional complexity.

Listed here are the per-year outcomes:

Check out the general numbers. One cause I just like the three-tier formulation as an outline of free agent spending is that utilizing a flat dollars-per-win estimate has directional bias. Should you used it to foretell all of the contracts in every bucket based mostly on their size and the participant’s projected WAR, it could miss low by about $10 million on the typical contract to a star, miss excessive by about $4 million on contracts to function gamers, and miss excessive by about $2 million on contracts to the 0-1 WAR bench choices.

One other manner of taking a look at it’s that if I compute confidence intervals for dollar-per-win charges for every tier independently, they’re appreciably completely different. Gamers within the 0-1 WAR tier have a $6.3-$8.6 million Ninety fifth-percentile confidence interval. The 1-2 WAR tier checks in at $7.8-$9.4 million, perhaps a small overlap with the underside tier. However the stars, the fellows within the high bucket? They verify in at a $10.6-$12.9 million estimate. They’re very clearly in a special class than the fellows on the backside of the market, even after accounting for variance and the inherent limits of pattern dimension.

I wished just a little extra statistical assurance, so I ran two extra assessments. First, I ran a permutation check. As an alternative of splitting gamers up by projected WAR in yr one, I break up them up utterly at random, 1,000 completely different occasions, after which checked whether or not my random splits did higher or worse than the by-WAR tiering. If grouping by WAR didn’t add any info, we’d count on to see a null end result, however as a substitute, solely one of many 1,000 random permutations grouped free brokers higher, by way of {dollars} per projected WAR, than the WAR-based kind. In different phrases, it’s impossible to be random.

Including shifting elements to a mannequin all the time makes it match higher, even when these shifting elements are simply modeling noise. To account for that, I ran a probability ratio check, which you’ll be able to consider as evaluating how good every of those strategies is at describing the contracts within the dataset, whereas penalizing added complexity in a elaborate math manner that I needed to examine. Should you spent a while taking stats in faculty, it’s like an F-test with a bit extra flexibility. I used that added flexibility to check the three-tiered mannequin in opposition to the one-tiered mannequin utilizing each a standard and lognormal distribution, and each distribution choices lowered the usual error of predictions by about 13% even after the complexity penalty. In different phrases, the three-tiered mannequin is displaying actual market differentiation.

Right here’s an fascinating impact, and one in line with my qualitative expertise of free company: The hole between the underside and high tiers is widening. From 2020 by 2023, gamers within the 2-plus WAR tier received about 40% extra per projected win than gamers within the 0-1 WAR tier. Even should you toss out the weirdo 2020 season, they received about 55% extra. Prior to now three years, they’ve gotten roughly double the {dollars} per projected win. In different phrases, the market is assigning increasingly per-WAR worth to high gamers. I can’t let you know whether or not that’s pushed extra by the provision aspect or the demand aspect of the equation (and make no mistake, each are drivers), however regardless of the trigger, star gamers who attain free company are getting paid extra on a relative foundation than ever earlier than.

That’s the conclusion. Listed here are a few of my takeaways about it. First, this tracks with how I take into consideration the perfect groups within the sport buying expertise, and I’m unsurprised to see the final route of the development. Observing the commerce market lately has led me to think about worth non-linearly for the Commerce Worth Sequence. You’ll be able to’t simply worth all wins equally and add them up; groups more and more worth stars at the next a number of than function gamers. Contracts signed earlier than the Decide/Ohtani/Soto mega-expansion, a rising tide that lifted all boats, are additionally beginning to look fairly engaging relative to new offers.

That segues properly into my subsequent level, a enjoyable little mathematical statement. I adjusted these numbers for inflation, and but a win above substitute value $9.8 million in 2020-2023 free agent contract negotiations, and $11.7 million from 2024-2026. The place’d the inflation adjustment go?! However I adjusted the numbers based mostly on complete league-wide payroll. Take 2021, for instance: Groups shell-shocked by COVID supplied comparatively small offers in free company, however current contracts meant that general league payrolls fell by a lot much less. Generally, although, free agent salaries are rising quicker than the remainder of the league, and by an enormous margin. In actual fact, it’s fairer to say that salaries for high free brokers are rising, as a result of the dollar-per-WAR values within the 0-1 WAR and 1-2 WAR buckets are maintaining with league-wide payroll inflation, whereas the dollar-per-WAR values going to stars are up 30% on a relative foundation lately. Sure, main league salaries are rising steadily, however it’s a stratified phenomenon.

Lastly, a fast observe on adjusting for the league minimal wage. Once I ran this evaluation the primary time, I break up it down into 0.5-WAR buckets to search for bizarre results. The celebs nonetheless made probably the most cash per projected win, however the 0.0-0.5 WAR bucket was a detailed second. I figured that one out fairly shortly, although: If a workforce pays a man $800,000 when the league minimal is $750,000, and he tasks for 0.1 WAR, that’s $8 million per WAR and likewise not a mirrored image of the tradeoff the workforce faces. To them, he didn’t value a ton greater than $50,000 (plus the lack of a versatile roster spot). Contracts in vary of the minimal wage, and in vary of substitute stage, would possibly fairly fairly behave in another way in recognition of that impact, so I included it in my modeling. I don’t assume it had an enormous impact apart from cleansing up that unusual anomaly.

I’m undecided how a lot bearing this analysis has on the way forward for the sport, as a result of I believe that the upcoming CBA has the potential to alter participant compensation, maybe seismically. Whether or not it’s a cap/flooring system (one thing I discover unlikely, to be clear), a revamped arbitration course of, or one thing I haven’t even considered but, the economics of the game might look fairly completely different the following time pitchers and catchers report back to spring coaching. However for now, the information leads me to a transparent conclusion: Groups worth stars non-linearly, and so they pay extra for high gamers than you’d assume from a broad take a look at the full market.