01
⚙️
12 Feature Inputs
Level proximity, age vs. league avg, 30-day performance stats, 40-man status, org opportunity, options remaining, position scarcity, performance trend, spring signal, prospect tier, and Rule 5 flag. All sourced from public data at evaluation date.
Stage 1–2
02
📊
CUS Score 0–100
All feature scores are summed and normalized to a 0–100 Composite Upward Score. No post-hoc weighting. Weights are baked into defined scoring ranges. Immutable at run date.
Stage 3
03
🎯
CUP Probability
CUS maps to a base call-up probability. Three sequential multipliers are then applied: Org Coefficient (30 orgs), Positional Base Rate, and Service Time Suppression — the most impactful force in the model.
Stage 3
04
📅
ECD Engine
CUP converts to an Estimated Call-Up Date via Pathway A (service time, calendar-driven, ±7d), B (injury/opportunity, ±21d), or C (performance unlock, ±45d). Each carries a confidence tier.
Stage 4 · v0.4