Dollars per WAR is a market average, not a price
Every winter the same arithmetic shows up in free-agent coverage. A player is projected for some number of wins above replacement, the projection is multiplied by the going rate for a win, and the product is declared his fair value. If the contract comes in above that number the team overpaid; below it, the team got a bargain. The commonly cited figure has risen over time, and the method is genuinely useful as a sanity check. But the number at its center is not a price that any team pays for any particular win. It is an average, computed after the fact, across a market full of buyers who value wins very differently.
The calculation works backward. Take the free-agent contracts signed in a period, estimate the WAR each player was projected to produce over the deal, and divide total dollars by total projected wins. Variations adjust for inflation, discount future seasons, or fit a regression rather than a simple ratio, but the logic is the same: the figure describes what the market as a whole spent per expected win. It says nothing about what a win was worth to the team that bought it, and nothing about what next year’s market will charge. It is a summary of past transactions, and like any summary it flattens the variation that matters most.
A win is not worth the same to everyone
The deepest problem is that the value of a win is nonlinear. Revenue from a season does not rise smoothly with victories; it jumps around the playoff line, where postseason gate, broadcast exposure, and the attendance and sponsorship bump of a contending summer all arrive at once. A team projected to finish near the cutoff gains far more from its marginal wins than a team projected to lose ninety games or one projected to cruise to a division title. For the rebuilder, an extra win mostly worsens its draft position. For the runaway favorite, it changes little about October.
That means the same player, with the same projection, is worth very different amounts depending on who signs him. Contenders near the line can rationally pay well above the market average and still come out ahead, while rebuilding teams can pass on a deal that looks cheap by the rule of thumb and be making the right call. Grading every contract against one league-wide rate treats a buyer’s situation as irrelevant, when it is often the whole story. The rate also cannot see wins concentrated in a single position of need, where a modest upgrade over a replacement-level hole can move a team’s odds more than a larger upgrade somewhere it already has depth.
Contract length and the aging curve
Free agents usually reach the market after their peak years, so a long deal buys a declining stream of wins. A fair-value calculation has to project each season separately, applying an aging curve that typically subtracts value year by year, and the shape of that curve is itself an estimate with real disagreement around it. Players with different skill profiles age differently: speed and defensive range tend to erode before plate discipline, and power can hold up longer than contact skills in some hitters. A single decline assumption applied to everyone misprices both groups.
Length also changes what the dollar figure means. Front-loaded or back-loaded structures, deferred money, and opt-outs shift when payments occur and who carries the risk, so two contracts with the same headline total are not the same commitment. The later years of a long deal are where the projection is weakest and the decline steepest, and they are also where a contender is often paying for wins it has already collected, having accepted the tail as the cost of getting the player during his useful seasons. Judging the whole deal by an average per-win rate misses that the team may have priced the early years and simply absorbed the rest.
Which WAR, and how certain
Wins above replacement is not a single statistic. Public versions differ in how they measure defense, how they handle pitchers, and where they set the replacement baseline, and the gaps are not trivial. A pitcher whose results outran his peripherals can look like a substantially different player depending on whether the system credits runs allowed or the components that predict them. A defender graded by one fielding model may look average by another. Any dollars-per-win figure inherits the version it was built on, so applying a rate derived from one system to a projection built on another quietly mixes units.
Even within one system, a projection is a distribution rather than a point. A player forecast for three wins might reasonably produce one or five, and injury risk fattens the bottom of that range for older players and pitchers. Two players with the same median projection can carry very different spreads, and a team near the playoff line may reasonably prefer the steadier one while a club needing a long shot to contend prefers variance. Multiplying a point estimate by a point rate produces a confident-looking number that hides all of this. A fair-value figure without a range attached overstates how much anyone actually knows.
The cheapest wins are not in free agency
The rate is built only from free-agent contracts, which are the most expensive source of wins in the sport. Players in their pre-arbitration years are paid near the league minimum regardless of how productive they are, and arbitration salaries, while rising, still trail what the open market would pay. Much of a successful team’s value comes from that surplus. Treating the free-agent rate as the price of a win ignores that teams can acquire wins far more cheaply through development and trades for controllable players, and that the real cost of a young player in a trade is the surplus he would have produced, not his salary.
This also explains why the market rate can look irrational. Teams rich in cheap young talent have room to spend heavily on the few positions they cannot fill internally, and they will pay a premium because their blended cost per win remains low. Teams without that pipeline face the opposite math. A contract that looks like an overpay in isolation can be a sensible allocation within a roster built mostly on surplus.
Payroll context and the luxury tax
The final distortion is that a dollar is not a dollar to every club. Revenues differ enormously across markets, so the opportunity cost of the same salary varies. And for teams above the competitive balance tax thresholds, each additional dollar of payroll carries a surcharge that escalates for repeat offenders and higher tiers, along with possible draft pick consequences. For those clubs, the effective cost of a free agent is meaningfully higher than his salary. For a team well below the threshold, there is no such penalty, and the sticker price is closer to the true cost.
Budget constraints matter in the other direction too. A team with a fixed payroll ceiling is not choosing between paying the rate and saving money; it is choosing among uses for a limited pool. Under that constraint, the relevant question is which allocation yields the most wins for the dollars available, which can make a player valued above the market rate a poor choice if the same money could fill two holes.
The honest read
Dollars per WAR is a reasonable starting point for asking whether a contract is in the right neighborhood, and the research that produced it made free-agent analysis far more disciplined than the era of judging deals by batting average and reputation. It is still a backward-looking market average. It blends buyers with very different stakes, ignores that wins near the playoff line are worth more than wins elsewhere, leans on aging assumptions and a particular version of WAR, excludes the cheap wins that drive most good rosters, and overlooks tax and budget context. Use it to frame a conversation, attach a range to any figure built on it, and ask what a win is worth to the team that signed the check before calling any deal a mistake.