The save is a structurally broken stat
The save was invented in 1969 by a Chicago sportswriter named Jerome Holtzman, who wanted a stat that would give relief pitchers some recognition. The intent was reasonable. The execution has been one of the most expensive accounting errors in baseball history. The save rule, as written, rewards the wrong inning, the wrong situation, and the wrong type of pitcher, and it has reshaped how every modern bullpen gets deployed in ways that demonstrably cost teams wins. The save is not a stat that's slightly miscalibrated. It is a stat whose structural design is incompatible with the situations it claims to be measuring.
What the rule actually requires
Under Rule 10.20, a save is awarded to a relief pitcher who finishes a game won by his team, is not the winning pitcher, and meets one of three conditions: he enters the game with a lead of no more than three runs and pitches at least one inning; he enters with the tying run on base, at bat, or on deck; or he pitches at least three effective innings. The first of these is the one that does almost all the work. A pitcher who enters the ninth inning with a three-run lead and no one on base, retires three batters in order — against the bottom of the order, often facing the worst hitters available — collects a save. The expected win probability for the team handing him the ball is roughly 97%. He converted a near-certain win into a certain win and got the same line on his stat sheet as a pitcher who entered with the bases loaded and the tying run on third in the eighth.
The two situations are not comparable. The first is essentially a formality. The second is a pitcher genuinely keeping his team in a game. The save rule treats them identically, which means the save total a closer accumulates at the end of the year is a count of how often his manager gave him the ninth with a comfortable lead, not a count of high-leverage situations he actually navigated.
The leverage-inversion problem
The structural error of the save is that the highest-leverage moments in most baseball games occur before the ninth inning. A bases-loaded, one-out, tied-game situation in the seventh is far more decisive than a clean ninth-inning save situation with a two-run lead. The pitcher who navigates the seventh- inning jam often gives way to a setup man, who gives way to a closer, who collects the save by handling the easier inning. The closer's stat line records the highest visible accolade for the lowest-leverage work, and the seventh-inning pitcher who actually saved the game gets a hold — a stat the market values at roughly one-tenth what it values a save.
Win probability data has documented this leverage inversion clearly for over twenty years. The average save situation has a leverage index around 1.5-1.8. The average seventh-inning appearance with men on base has a leverage index of 2.5 or higher. Managers who deploy their best reliever exclusively in save situations are using their best pitcher in the lower- leverage spot the majority of the time. The stat that defines the role is the stat that is responsible for the misuse.
The "closer mentality" reinforcement loop
Once a pitcher accumulates saves, the market re-evaluates him as a closer, and his contract reflects that valuation. The gap between closer pay and setup-man pay is enormous — often 3-5x for pitchers with similar underlying skill. The pay gap is driven by the save total, which is driven by bullpen deployment, which is driven by the perceived importance of the closer role, which is driven by the prominence of the save. The entire loop is downstream of an accounting rule.
This is why a 35-year-old closer with declining velocity but a string of saves on a contending team gets a three-year contract with a high salary, while a 27-year-old setup man with better strikeout and walk rates gets a one-year deal at a fraction of the price. Both pitchers have similar future expected production. The market is paying for the save history, which was produced by manager deployment, which was shaped by the rule that defines the save in the first place.
The blown save asymmetry
The save also creates a perverse incentive on the negative side. A pitcher who enters a tied game in the ninth and gives up a run takes a loss but not a blown save. A pitcher who enters with a one-run lead and gives up a run takes a blown save, which is treated by the market as a much more serious failure than a regular loss. The two outcomes are statistically identical — one earned run in one inning in a high- leverage spot — but the accounting treats them as different categories of failure with different career consequences.
This pushes managers toward deploying their closers only in save situations, because deploying them in tie games risks a loss without the upside of a save. The result is that the most leveraged tie-game situations in close games are handled by the team's second- or third-best reliever, while the best reliever sits in the bullpen waiting for a lead to materialize. The most-quoted bullpen stat is the structural cause of why the best bullpen arms are unavailable for the most important moments.
What replaces it for actual evaluation
The advanced bullpen metrics that have circulated since the early 2000s — leverage index, win probability added, FIP, expected ERA on Statcast inputs — are all better than the save at evaluating relief pitchers. WPA in particular captures what a closer is supposed to be doing: shifting the team's chance of winning by surviving high-leverage moments. A closer who accumulates +3 WPA across a season has genuinely added three wins of value. A closer who accumulates +0.8 WPA across a season with 40 saves has been used in low-leverage spots and added much less value than his save total suggests.
The teams that have moved away from the traditional closer model in favor of high-leverage deployment of their best reliever, regardless of inning, have generally been the same teams that have outperformed their roster on paper in the playoffs. The structural advantage isn't subtle. Using your best pitcher in the inning that matters most produces more wins than saving him for the inning where the save accrues.
How to read a bullpen stat line
A save total is fine as a record of how often a pitcher finished games his team led, and that's basically all it is. It doesn't tell you whether he's a good pitcher. It doesn't tell you whether his usage was efficient. It doesn't tell you whether the leverage of the innings he pitched justified the contract he's earning. For those questions, the relevant stats are leverage index, WPA, FIP, and strikeout-minus-walk rate. All of them are publicly available. None of them appear on the broadcast graphic that announces the closer entering the game.
The save isn't going away because too many contracts and too many traditions are built on top of it. But the gap between the stat and the underlying pitching value has been visible in the data for two decades, and the teams winning World Series have stopped treating the save total as evidence of bullpen quality. The teams that haven't keep paying premium prices for closer history that is mostly a record of how the save rule happens to be written.