The Infield Grid: Baseball as the Natural Home of Sports Analytics

On a frigid November evening in 2022, with the bases loaded, two outs, and the Houston Astros clinging to a one-run lead in the bottom of the ninth inning of World Series Game 6, right fielder Kyle Tucker launched himself backward toward the wall, his glove stretching over the outfield fence to snag a 104-mph line drive that 94% of Major League Baseball’s Statcast tracking system projected would fall for a hit. The catch sealed the Astros’ second championship in six years, but the postgame conversation split immediately along the fault line that has defined baseball for more than 150 years: the 72-year-old scout sitting behind the Astros’ dugout shook his head and told reporters, “I’ve watched 50 years of baseball. That ball was hittable. Tucker just got lucky.” Meanwhile, the team’s director of analytics pointed to the Statcast dashboard on his tablet and noted that the catch had a 6% catch probability, the lowest of any successful out recorded in the 2022 postseason, and that Tucker’s defensive runs saved (DRS) metric for the season ranked second among all American League outfielders.

This split reaction was not unique to the 2022 World Series. For as long as organized baseball has existed, two competing visions of how to evaluate talent and strategy have vied for dominance: one rooted in the subjective, experience-based intuition of scouts who watch players in person, and one rooted in the objective, replicable data generated by the sport’s unique structure. Unlike football, basketball, or soccer, where continuous, interdependent team flow makes it nearly impossible to isolate individual contribution, baseball is a series of discrete, turn-based interactions: a pitcher throws, a batter swings or lets the pitch go, a runner decides to steal or hold, a fielder fields a ball and throws to a base. Each of these interactions produces a clear, binary outcome that can be recorded, categorized, and aggregated over time to build a standardized dataset of player and team performance. It is this structural quirk—this natural alignment between the sport’s rules and the logic of quantitative analysis—that made baseball the inevitable first home of the sports data revolution, a story that begins not with a supercomputer or a multi-million dollar analytics department, but with a 19th-century journalist and a hand-drawn grid of numbers.

The Structural Quirk That Made Baseball Analytics Inevitable

To understand why baseball led the data revolution across all athletics, one must first understand its unique structural properties. Every other major team sport operates on a model of continuous, overlapping action: in basketball, 10 players move simultaneously across a 94-foot court, with passing, cutting, and screening happening in overlapping sequences that make it impossible to isolate the contribution of any single player to a made basket. In soccer, 22 players share a 100-yard pitch, with possession changing hands in fluid, unstructured sequences that rarely produce clear attribution for goals or defensive stops. Even American football, which is divided into discrete plays, relies on 11-player coordinated action for every snap, where rushing yards are often the product of blocking as much as running, and passing yards are as much a product of receiver route-running as quarterback accuracy.

Baseball, by contrast, is a game of isolated, sequential events. Each plate appearance is a self-contained interaction between a pitcher and a batter, with no other players able to directly influence the outcome of the at-bat[^1]. Each stolen base attempt is a one-on-one contest between a runner and a catcher, with no teammates able to block the runner’s path or alter the catcher’s throw. Each fielding play is a discrete test of a fielder’s ability to react to a batted ball and deliver it to the correct base, with no defensive teammates able to cover for a fielder’s error (unlike a soccer defender who can block a shot intended for a teammate, or a basketball player who can rotate to cover a teammate’s defensive mistake). This structure means that every play in baseball has a clear, binary outcome that can be recorded without subjective interpretation: a pitch is either a ball or a strike, a batted ball is either a hit or an out, a runner is either safe or out. There is no equivalent of a “no-call” in baseball, no “possession” ambiguity, no “offside” rule that requires subjective judgment of player positioning at the moment of a pass.

This structural clarity was not immediately obvious to the sport’s early practitioners, who viewed baseball as a test of physical skill and competitive grit rather than a series of quantifiable events. For the first 30 years of organized baseball, from the 1840s through the 1860s, record-keeping was haphazard and inconsistent: different teams used different definitions for hits, outs, and runs, and newspapers often printed conflicting scores for the same game, relying on telegraph reports that were frequently delayed or inaccurate. The turning point came in 1857, when the National Association of Base Ball Players (NABBP), the sport’s first governing body, published the first standardized set of scoring rules, creating a shared vocabulary for recording game events[^2]. But it was a young journalist named Henry Chadwick who turned these standardized rules into a replicable dataset that would lay the foundation for 150 years of baseball analytics.

Chadwick, a 24-year-old British immigrant working for the *New York Clipper* in 1859, had grown up watching cricket and was frustrated by the lack of consistent data about baseball games, which were rapidly growing in popularity across the northeastern United States. Drawing on his experience with cricket scorecards, Chadwick designed a standardized box score that mapped the nine defensive positions of baseball to a simple 3x3 grid, which he called the “infield grid.” Each cell in the grid corresponded to a specific defensive position: the top row represented the outfield (left, center, right), the middle row represented the infield (shortstop, second base, first base), and the bottom row represented the corners (third base, pitcher, catcher). Every hit, out, or error was marked in the cell corresponding to where the ball was hit or fielded, allowing readers to visualize a player’s hitting tendencies or a fielder’s range without ever watching the game in person.

Chadwick’s box score was revolutionary not just for its design, but for its replicability. For the first time, two different scorers recording the same game would produce identical box scores, eliminating the conflicting reports that had plagued baseball coverage for decades. By 1861, Chadwick’s box score had been adopted by most major New York newspapers, and by 1869, the first fully professional baseball team, the Cincinnati Red Stockings, used standardized box scores to track player performance across their 57-game undefeated season. The data generated by these box scores allowed the Red Stockings’ manager, Harry Wright, to identify which players were most consistent at getting on base and making plays, a radical departure from the era’s dominant scouting model, which relied on ex-players’ subjective assessments of a player’s “form” and “character.”

The formalization of baseball’s data infrastructure accelerated in 1876, when the newly formed National League (NL) made standardized box scores a mandatory requirement for all league games, with official scorers appointed by the league to ensure consistency[^3]. This created the first continuous, standardized dataset in professional sports: for the first time, teams could compare player performance across games, seasons, and even leagues, using the same definitions and metrics. By 1878, the first career batting average leaderboard was published, ranking players by their cumulative hits divided by at-bats, a metric that would remain the gold standard for evaluating hitters for the next 70 ye