Which Dodgers hitters project the strongest combination of 2027 availability and offensive performance, and how much does Statcast add beyond traditional batting statistics?

We separate:

• Playing time: probability of reaching at least 150 PA.
• Conditional performance: next-season wRC+ among hitters who reach 150 PA.

The final holdout is 2026; the application forecast is 2027. The decision focus is the Dodgers roster.

• FanGraphs all-player batting exports: 2015–2026.
• Baseball Savant season-level Statcast exports: 2015–2026.
• Chadwick Register: stable FanGraphs-to-MLBAM identifier mapping.

The 2015–2025 audit found complete eligible cross-source joins and no duplicate player-season keys. Primary non-2020 retention ranged from approximately 74% to 79%.

2020 is included descriptively but excluded from primary transitions because it was a shortened season with an unusual playing-time environment and tracking-system transition.

For Dodgers roster planning, 150 PA preserves a useful sample while avoiding the noisiest part-time population. A 300 PA screen would focus review on regular players, but it would exclude legitimate role-player forecasts.

Retention declines as the prior-season playing-time threshold increases. We keep 150 PA as the primary modeling threshold and use 300 PA only as a stricter review sensitivity.

The historical feature set uses PA, wRC+, wOBA, ISO, BB%, and K%. The augmented model adds xwOBA, hard-hit rate, barrel rate, sweet-spot rate, and average exit velocity.

Frozen choices

• Historical logistic model for playing-time eligibility.
• Historical + Statcast ridge model for conditional wRC+.
• Fixed XGBoost versions retained as sensitivity analyses.

All imputation and scaling occur inside the model pipeline.

The Statcast addition helps conditional wRC+ modestly. XGBoost does not provide enough improvement to replace the simpler ridge model.

The holdout visualization showed good broad calibration, with the largest conditional-performance errors among lower-PA players.

The selected models were refit through 2026 and applied to Dodgers hitters with at least 150 PA in 2026.

The strict review screen requires:

• Eligibility probability ≥ 0.50.
• Predicted wRC+ ≥ 110.
• Predicted wRC+ at least 10 points above 2026 wRC+.

The shortlist below is intended for Dodgers roster and player-development discussion, not as an automatic transaction recommendation.

Player labels identify the selected Dodgers hitters. The light green upper-right region marks players meeting both review thresholds: at least 50% eligibility probability and 110 predicted wRC+.

The 150 PA threshold is the eligibility target used throughout the project. The review screen emphasizes predicted wRC+ of at least 110 and an improvement of at least 10 points over 2026.

These are Dodgers roster-planning priorities, not automatic acquisition recommendations. Add health, roster status, defensive role, and expected playing time during baseball review.

• Only five expanding development folds were available.
• 2020 is not exchangeable with a normal season.
• Statcast missingness is non-random and flagged in the forecast output.
• No age, injury, contract, park, role, or roster-context variables are modeled.
• 2027 outcomes are not yet available.
• Forecasts are predictive and associative, not causal.

Project files:

• Complete ranked forecast table
• Dodgers forecast shortlist
• Forecast-building script
• Forecast pipeline tests
• Project requirements
• Complete project repository on GitHub