How baseball analytics works

Baseball is the sport where analytics grew up. Its play is broken into discrete, countable events — a pitch, a swing, an out — which made it the natural home for statistical analysis decades before player tracking existed anywhere else in sport. [[sabermetrics|Sabermetrics]], the field's own name for itself, now spans everything from a simple ratio you can compute by hand to pitch-by-pitch physics captured by stadium-mounted cameras and radar.

The questions analytics is built to answer

Front offices use it to answer three recurring questions: is a player's performance sustainable or due for regression, how much of a pitcher's results are their own doing versus their defense's, and which of two similar-looking hitters or pitchers is actually the better bet going forward. Broadcasters and fans use a lighter version of the same tools to explain why a team is over- or under-performing its record.

The data it runs on

Baseball analytics runs on three tiers of data that arrived in roughly this order historically:

• Box score and play-by-play data — hits, walks, strikeouts, outs, base-runner states — has existed since the sport's earliest recorded seasons and remains the foundation for most traditional and sabermetric statistics alike.
• Batted-ball data, recorded by stadium tracking systems, captures the physical outcome of contact: exit velocity (how fast the ball left the bat) and launch angle (the vertical angle it left at). These two numbers together predict a batted ball's likely outcome — a hard-hit ball at an optimal launch angle is a probable extra-base hit regardless of where a fielder happened to be standing — far better than the outcome alone, because outcome is partly luck.
• Pitch-level data captures velocity, spin rate and movement on every individual pitch, which underlies modern pitching evaluation and pitch-design work that didn't exist as a discipline before this data was available.

Measuring hitting: beyond batting average

Batting average counts hits per at-bat and ignores walks and power entirely, which is why it fell out of favor as the primary hitting metric. [[on-base-plus-slugging|On-base plus slugging]] (OPS) adds on-base percentage (which credits walks) to slugging percentage (which credits extra bases), giving a single number that rewards both getting on base and hitting for power: OPS = OBP + SLG. It is not a perfectly weighted combination — a point of OBP and a point of SLG are not equally valuable — but it is a large, well-understood improvement on batting average and remains the most widely quoted advanced hitting stat.

[[batting-average-on-balls-in-play|Batting average on balls in play]] (BABIP) measures how often a ball put in play (excluding home runs and strikeouts) falls for a hit. It is used less as a hitting-skill metric and more as a diagnostic: a hitter or pitcher with a BABIP far outside their career norm is often experiencing unusual luck — good or bad fielding positioning, ball placement, or plain variance — that is likely to fade, rather than a genuine change in ability.

Measuring pitching: separating a pitcher from their defense

A pitcher's earned run average is shaped heavily by the defense playing behind them and by the ballpark they pitch in, neither of which is the pitcher's doing. [[fielding-independent-pitching|Fielding independent pitching]] (FIP) isolates the outcomes a pitcher controls most directly — strikeouts, walks, hit-by-pitches and home runs allowed — and converts them to an ERA-like scale, deliberately excluding balls put in play, which depend heavily on the fielders behind the pitcher. A pitcher with an ERA well above their FIP has often been hurt by weak defensive support or bad luck on balls in play, and is a common target for a team betting on regression toward that FIP in the following season.

How the work is done in practice

A modern front office blends these layers rather than picking one. A trade evaluation typically starts with outcome-based stats (OPS, FIP) to screen for production, checks batted-ball data (exit velocity, launch angle) to see whether that production reflects genuinely hard, well-angled contact or a run of fortunate placement, and checks BABIP against career norms to flag anyone whose recent numbers look like luck rather than skill. Player development departments use pitch-level data the same way engineers use telemetry: identifying which specific pitch, grip or release point change would most improve a pitcher's results, an approach that has visibly changed how pitching is coached over the past decade.

At the team level, [[pythagorean-expectation|Pythagorean expectation]] — originally developed for baseball before spreading to other sports — estimates the win total a team's run differential (runs scored versus runs allowed) implies, independent of its actual win-loss record. A team significantly outperforming its Pythagorean expectation is often winning an unusual number of close games, a pattern that tends not to repeat, which is one reason analysts treat such a team's record with more caution than its raw standings position suggests.

Common mistakes and misreadings

• Trusting early-season stats. Batting average, ERA and most rate stats need hundreds of plate appearances or dozens of innings before they stabilize; a hot or cold first month is usually noise.
• Treating BABIP as purely a luck indicator. Genuinely hard-hit, well-placed contact does sustain a higher BABIP for some hitters — the number needs to be read alongside exit velocity and launch angle, not in isolation.
• Using OPS to compare hitters across very different ballparks or eras without adjusting for park and league scoring environment, both of which shift the baseline significantly.
• Assuming FIP is a complete picture of a pitcher's skill. It deliberately ignores what happens on balls in play, which means it can undervalue pitchers who are genuinely good at inducing weak contact and limiting hard-hit balls.
• Reading a single season's Pythagorean gap as a permanent talent gap, when a meaningful part of it is typically variance in close-game outcomes that regresses the following year.

For how front offices weigh these numbers alongside scouting and injury-risk data when building a roster, see how front offices use analytics. Browse every tool in this category for performance-data vendors, most of which cover baseball alongside other sports.