Baseball's Statistical Revolution Has Gone Too Far
Sabermetrics rescued baseball from a culture of bad information. Then it went overboard, and today fans can't follow it.
I was a sabermetrics evangelist before most people knew what sabermetrics was. 25 years ago I was reading old copies of Bill James's Baseball Abstracts, the self-published annual volumes in which James, a night security guard in Kansas, was systematically dismantling everything conventional baseball wisdom thought it knew about the game. His 1983 Abstract in particular was, to a certain kind of baseball-obsessed reader, a revelation. Here was someone applying genuine empirical rigor to a sport that had been run on gut instinct, tobacco-stained intuition and the self-serving mythology of old men who had played the game and confused having played it with understanding it.
The conventional statistics of that era were, by any honest assessment, terrible. Batting average told you how often a player got a hit but said nothing about walks, which meant it systematically undervalued one of the most important offensive skills in the game. RBIs told you how many runners happened to be on base when a player came to the plate as much as they told you anything about the player himself. Saves told you how many times a closer had entered a game with a lead of three runs or fewer, rewarding ordinary pitching performances while masking instances of failure. The statistics on the scoreboard and the back of the baseball card were actively misleading and the men who ran baseball teams treated them as gospel.
Sabermetrics argued that on base percentage matters more than batting average because getting on base by any means is what drives run scoring. Slugging percentage captures power more accurately than home run totals alone. Stolen bases are overrated because nobody was properly accounting for the cost of getting caught. Lineup order matters far less than managers believed. These were not radical claims. They were conclusions that followed from actually looking at the data. The resistance to them from inside baseball was, for years, almost total. When I worked briefly in a Major League Baseball player development office in the mid 2000s, the advanced metrics I was running were treated with a mixture of curiosity and condescension. The presumption was that people who understood baseball in their bones did not need statistics to tell them what they already knew. The statistics people were nerds compensating for their lack of feel for the game.
That attitude was wrong, and the teams that abandoned it first proved it conclusively. The early 2000s Oakland Athletics, working with a payroll a fraction of the Yankees, competed for division titles by systematically acquiring players the market undervalued because conventional statistics failed to capture what made them good. The 2004 Boston Red Sox ended an 86-year championship drought in part by applying the same thinking. The revolution was real and it was correct and for a long time I would have argued for it against anyone.
I am not sure I would make that argument today.
Where the Revolution Went
The problem is not that baseball teams started using better statistics. The problem is that the statistical arms race that followed has produced metrics so abstracted from anything a fan can observe, calculate, or intuitively connect to the game they are watching that it has effectively cleaved baseball into two separate experiences: the one happening in the stands and on the field, and the one happening in the analytics department, with very little communication between them.
WAR, Wins Above Replacement, is the canonical example. It is now the dominant currency of baseball evaluation, the number reached for first in any argument about a player's value. And it is, by design, not calculable by a fan watching a game. Its formula is not publicly available in any complete form. It incorporates defensive metrics that are themselves composites of other metrics, adjustments for park factors, positional adjustments and replacement level baselines that require their own explanation. A fan sitting in the stands cannot look at what a shortstop just did and form any intuition about how it moved his WAR. The statistic is real and captures something meaningful. It is also opaque to the people who are supposed to care most about the game.
This matters because the statistics fans can follow are the statistics that create the narratives fans invest in. The home run chase works as a story because everyone understands what a home run is and everyone can watch the number climb. The race to .300 batting average works because people implicitly understand what a hit is and the calculation is simple enough to do in your head. Stolen base records are interesting because you can watch a player steal a base. And the number of how many times he has done that is right there on the scoreboard and its meaning is self-evident. These narratives drove fan engagement for generations not because the statistics behind them were the most analytically rigorous measures of player value but because they were legible, and legibility is what allows a fan to have a stake in what they are watching.
Teams no longer care about batting average, which means broadcasters cover it less, which means the race to .300 has quietly ceased to exist as a storyline. Teams pull starting pitchers before they complete seven innings as a matter of analytical routine, which means no-hitters and perfect games are increasingly collaborative efforts that feel less like individual achievements. Every time a starter is pulled before getting to even try for a generational feat is an opportunity at a lifetime memory robbed from the fans, and for what? A closer in 2025 is deployed according to leverage index models rather than in the ninth inning as a matter of course, which means the save, whatever its statistical limitations, no longer reliably describes what the best relievers actually do. One by one, the narrative hooks that gave fans something to track and argue about have been quietly retired by front offices optimizing for outcomes the fans in the seats cannot see or understand. The presumption that winning the championship is the only thing that matters belies the infinite additional potential storylines that might emerge both within a game and within a season. Pete Rose, Barry Bonds and Mark McGwire all set single-season and all-time records in moments that are indelible in the lore of baseball history for teams that did not even win their divisions. Even a perfect game lost in the 9th inning, like Ron Robinson's in 1988, is a story that floats around a hot stove baseball discussion decades later.
I remember 2002, when Vladimir Guerrero and Alfonso Soriano were both chasing 40 home runs and 40 stolen bases simultaneously. It wasn't the most urgent or critical set of records, but it helped the sport feel alive with a specific kind of drama that required no explanation and no spreadsheet. Chases like those are so much more rare now, not because the players are less talented but because the organizational incentives that would produce such chases and moments have been analytically engineered away for the sake of winning in an almost mechanical fashion. Stolen bases were undervalued for decades and sabermetrics correctly identified that. Then sabermetrics over-corrected and the stolen base as a narrative device essentially vanished from the sport for the better part of a decade, until the pitch clock reforms brought it partially back.
The Arms Race Problem
There is also an uncomfortable irony at the heart of where the statistical revolution has landed. One of the original arguments for sabermetrics was democratic: small market teams could compete with large market teams by being smarter rather than richer. The early Oakland Athletics were the proof of concept. If you found the market inefficiencies, the undervalued players and the metrics your opponents were ignoring, you could build a competitive roster without a New York payroll.
That argument has largely collapsed. The sophisticated analytics departments that small market teams pioneered have been replicated and surpassed by large market teams with the resources to hire entire floors of ex-McKinsey data scientists, build proprietary tracking systems and run analyses that smaller organizations simply cannot afford to match. The arms race that sabermetrics started has, in the long run, advantaged the teams with the most resources to wage it. The revolution that was supposed to democratize competition has instead produced a new kind of arms race that the Yankees and Dodgers are better equipped to win than the Rays and Athletics ever were.
What Baseball Needs Back
None of this means the statistical revolution was a mistake. On base percentage really does matter more than batting average. Outs really are precious. Relief pitcher usage really was inefficient for decades and the analytics correction was real and warranted. I would not trade the era of better information for the era of RBIs and pitcher wins as the measures of excellence.
But there is a difference between using better statistics to make better decisions and allowing the statistical apparatus to become so complex and so proprietary that it alienates the fans who are supposed to be the point of the enterprise. Baseball is not a hedge fund. It is not a NASA launch sequence. It is a game that people are supposed to watch and feel something about, and feeling something requires understanding enough of what is happening to have a stake in it.
The sport needs statistics that live on the scoreboard as well as in the front office. It needs narratives that fans can follow with their eyes and their arithmetic rather than only with a subscription to a data platform. It needs to remember that the home run chase and the stolen base record and the race to .300 were not just quaint artifacts of a less sophisticated era. They were the mechanisms by which millions of people stayed emotionally connected to the sport across a 162-game season, and dismantling them in the name of optimization has cost something real.
I spent years arguing that baseball needed to get smarter about its numbers. It did. Now it needs to get smarter about which numbers belong to the fans.
