Baseball’s analytical landscape has been reshaped by the arrival of Statcast, the league‑wide tracking system that captures every swing, pitch and fielding motion in three‑dimensional detail. Those raw data streams feed into a family of ‘expected’ statistics, which strip away the noise of sequencing and defense to estimate what a player’s performance should look like if luck were removed from the equation.
One of the most widely used projection tools that leans on those expectations is ZiPS, a system created by Dan Szymborski that blends spray charts, plate‑discipline metrics and historical comparables to forecast a player’s future output. ZiPS does not simply project raw totals; it adjusts for factors such as home‑run estimators, regression toward the mean and the typical volatility of small‑sample outcomes, producing a nuanced picture that often diverges from traditional counting stats.
The model’s predictions have already sparked debate, especially when they flag players who are outperforming or underperforming relative to their expected metrics. In some cases the discrepancies are subtle, in others they are stark enough to influence contract talks and roster decisions.
The Human Element Behind the Numbers
Take Byron Buxton, whose elite bat speed and hard‑hit rate have not yet translated into the home‑run totals that ZiPS expects, while Otto Lopez’s .310 batting average and unusually high BABIP have earned him an MVP‑type narrative that the model attributes largely to luck. Similarly, Mickey Moniak’s elevated BABIP and projected home‑run rate suggest a breakout that ZiPS believes is plausible, whereas Jackson Merrill’s recent dip in production appears inconsistent with improvements in bat speed and hard‑hit percentage.
Veteran hitters like Will Smith and Alex Bregman are also on the radar; ZiPS thinks Smith’s current stat line is lagging behind his expected slugging, while Bregman’s power surge may be overstated by the model’s home‑run estimator. The system’s adjustments are not limited to hitters. Pitchers such as Luis García Jr. and Kyle Schwarber see their projected home‑run totals trimmed in favor of more sustainable patterns of contact, while Vladimir Guerrero Jr. and Fernando Tatis Jr. are flagged as underperforming in the power department despite eye‑catching swing metrics.
These examples illustrate how ZiPS uses a blend of data‑driven regression and historical benchmarks to separate genuine skill from fleeting variance, offering a compass for teams navigating free‑agency markets and in‑season roster moves. Behind every projection lies a story of preparation, injury history and scouting nuance that numbers alone cannot capture. Analysts still rely on gut instinct, clubhouse observations and the occasional anecdote from coaches to interpret why a model might flag a player as an over‑ or under‑achiever.
As Statcast continues to expand its sensor network and as machine‑learning techniques become more sophisticated, the gap between expected and actual performance is likely to narrow. Yet the very act of quantifying ‘expected’ outcomes keeps the conversation alive, reminding fans and front offices that baseball remains a game of both measurable skill and unquantifiable heart.