What Metrics Matter Most in MLB Predictions

The Problem with Traditional Statistics

Traditional baseball statistics like batting average, ERA, and RBIs have been around for over a century. They are familiar and easy to understand. But familiarity does not mean usefulness for prediction. Many traditional stats include significant noise that makes them unreliable indicators of future performance.

Consider batting average. A hitter who goes 3-for-10 has a .300 average. A hitter who goes 2-for-10 has a .200 average. The difference between these outcomes could easily come down to a few feet on a line drive, or an inch on a ground ball that either finds a hole or gets snagged. Batting average treats all hits the same and ignores walks entirely. It captures results but obscures the underlying skill.

ERA, or Earned Run Average, has similar problems. A pitcher who allows three runs in six innings has a 4.50 ERA for that start. But those three runs might have come on a bloop single, a weakly hit ground ball that found a hole, and a fly ball that carried just over the fence on a windy day. Or they might have come on three towering home runs off hanging breaking balls. ERA treats these scenarios identically, even though one suggests bad luck and the other suggests poor performance.

What Makes a Statistic Predictive

A predictive statistic is one that correlates strongly with future performance. The key question is not whether a stat describes the past accurately, but whether it tells you something reliable about what comes next.

The most predictive statistics tend to share certain characteristics. They isolate what a player can control. They remove or minimize the impact of random variation. They stabilize quickly, meaning they become reliable indicators with relatively small sample sizes. And they focus on process rather than results.

Stabilization and Sample Size

One of the most important concepts in baseball analytics is stabilization. Different statistics require different amounts of data before they become reliable. Strikeout rate for pitchers stabilizes relatively quickly, often within 100 batters faced. Home run rate takes much longer. Understanding these timelines helps avoid overreacting to small samples.

For hitters, plate discipline metrics like walk rate and chase rate stabilize faster than outcome-based stats like batting average or slugging percentage. For pitchers, command metrics and strikeout-to-walk ratio become reliable sooner than ERA or WHIP. This is why predictive systems often lean heavily on the faster-stabilizing metrics, especially early in a season.

Pitching Metrics That Predict Performance

For starting pitcher evaluation, several metrics offer more predictive value than traditional ERA.

FIP: Fielding Independent Pitching

FIP measures what a pitcher's ERA would look like if they had league-average defense and luck on balls in play. It focuses only on outcomes the pitcher directly controls: strikeouts, walks, hit batters, and home runs. FIP is generally more predictive of future ERA than past ERA itself, because it strips out the noise from defense and sequencing.

xFIP: Expected Fielding Independent Pitching

xFIP goes one step further by normalizing home run rate to league average. The logic is that home run per fly ball rate is highly variable and tends to regress toward the mean. A pitcher who has allowed an unusually high or low percentage of fly balls to leave the yard is likely to see that rate move back toward normal. xFIP anticipates this regression.

SIERA: Skill-Interactive ERA

SIERA is the most sophisticated of the commonly used predictive pitching metrics. It accounts for how different types of batted balls interact, recognizing that ground ball pitchers and fly ball pitchers have different expected outcomes. SIERA tends to be the most predictive of future ERA among the widely available metrics.

Strikeout and Walk Rates

The most fundamental predictive pitching stats are strikeout rate and walk rate. These outcomes are almost entirely within the pitcher's control. A pitcher with elite strikeout numbers and low walk totals is likely to continue performing well. A pitcher with poor command who is posting a low ERA is a regression candidate.

For a complete guide to pitcher evaluation before games, see How Starting Pitchers Are Evaluated Before a Game.

Offensive Metrics That Signal Future Performance

On the hitting side, a similar logic applies. Metrics that focus on process and skill tend to be more predictive than those that simply track results.

wOBA: Weighted On-Base Average

wOBA weights each offensive outcome by its actual run value. A home run is worth more than a double, which is worth more than a single, which is worth more than a walk. Unlike batting average or on-base percentage, wOBA captures the full value of what a hitter produces. It is more predictive than traditional stats because it accounts for the quality of contact, not just whether contact resulted in a hit.

wRC+: Weighted Runs Created Plus

wRC+ is wOBA adjusted for park and league, then scaled so that 100 is league average. A hitter with a 120 wRC+ is 20% better than average. This adjustment matters because hitting in Coors Field is not the same as hitting in Petco Park. wRC+ puts everyone on the same scale, making comparisons more meaningful.

Exit Velocity and Barrel Rate

Statcast data has introduced new ways to evaluate hitters. Exit velocity measures how hard a ball is hit. Barrel rate measures how often a hitter produces optimal contact. Both metrics are highly predictive because they measure the quality of contact regardless of where the ball lands. A hitter with elite exit velocity and barrel rate is likely producing hard contact that will eventually turn into hits and extra-base hits, even if the current results are disappointing.

Plate Discipline Metrics

Walk rate, strikeout rate, and chase rate all provide insight into a hitter's approach and discipline. These metrics stabilize faster than batting average and correlate with offensive success. A hitter who rarely chases pitches outside the zone and makes consistent contact is likely to maintain production over time.

Contextual Adjustments

Raw numbers only tell part of the story. Context matters significantly in baseball evaluation.

Park Factors

Different stadiums affect offense in different ways. Coors Field in Denver dramatically inflates run scoring due to altitude. Petco Park in San Diego suppresses it. Fenway Park's Green Monster creates unique conditions for hitters and pitchers alike. Any serious evaluation must account for where games are played.

Opponent Quality

A pitcher who has faced a weak schedule will have better raw numbers than one who has faced the league's best offenses. Similarly, a hitter who has seen a difficult run of opposing pitchers may have depressed statistics that understate true ability. Adjusting for opponent quality helps identify which performances are sustainable and which reflect circumstance.

Platoon Splits

Most hitters perform differently against left-handed and right-handed pitching. Some pitchers have dramatic platoon splits as well. Understanding these splits is essential for predicting performance in specific matchups. A hitter who mashes left-handed pitching but struggles against right-handers is not the same player in every game.

Why This Framework Matters for Prediction

The distinction between descriptive and predictive statistics is not academic. It has practical implications for anyone trying to anticipate baseball outcomes.

A team that is outperforming its run differential is likely to regress. A pitcher with a low ERA but poor peripherals is due for trouble. A hitter with elite exit velocity but a low batting average is probably getting unlucky. These patterns emerge from understanding which metrics are predictive and which are noisy.

None of this guarantees correct predictions. Baseball contains irreducible randomness that no metric can eliminate. But using the right statistics improves the probability of being right over time. For more on the limits of prediction accuracy, see Can You Actually Predict Baseball Games Accurately.