The Projection Puzzle
When the ZiPS projection system was introduced, it quickly became a staple for forecasting major‑league performance. Yet over the past five seasons its forecasts for the Milwaukee Brewers have repeatedly lagged behind the team’s actual output, creating a persistent gap that has drawn the attention of analysts and fans alike.
The discrepancy is not an isolated anomaly. Historical data show that ZiPS has underperformed with the Brewers more than with any other franchise in the league, a pattern that suggests a structural bias rather than random variance.
A Persistent Undervaluation
One of the most elusive sources of error lies in the model’s handling of trade‑deadline activity. The timing, magnitude and even the existence of a move can dramatically reshape a club’s projected value, but the system struggles to incorporate those sudden shifts, leaving a blind spot that can skew entire season outlooks.
What sets the Brewers apart, however, is their willingness to reward over‑performance. Between 2021 and 2025, 81 percent of qualifying players who exceeded their projected wRC+ or ERA+ received more playing time than originally anticipated, a strategy that inflates real‑world results while the projection model remains static.
The Brewers’ Competitive Edge
The system’s caretaker, who has devoted years to refining ZiPS, acknowledges the challenge. The Brewers’ unique talent evaluation—often emphasizing tools and makeup over traditional statistics—poses a difficult puzzle for any algorithm that relies on historical comparables.
Addressing these shortcomings requires more than incremental tweaks. The author is now focused on uncovering systemic issues within the model, from the way it weights playing‑time adjustments to the assumptions built into its trade‑deadline logic, with the aim of producing a more faithful representation of the team’s true competitive standing.
Looking Forward
By dissecting the interplay between projection methodology and the Brewers’ unconventional player development, the effort seeks to close the gap between forecast and reality, ensuring that future models reflect the team’s distinctive strengths more accurately.