Call-Up Probability Engine · Season Complete · 107 Graded Outcomes

Every Number,
Graded.

Name: MKDC Baseball CUP Leaderboard: MLB Prospect Call-Up Probabilities
Creator: MKDC Baseball
Published: 2026-03-20
License: https://mkdcbaseball.com/methodology/

One question: when does a prospect get called up? In 2026 this model published 107 probabilities before the outcomes were known, then graded every one, misses included.

See the Record How it works

One email a week. Free. Unsubscribe anytime.

107

Graded Outcomes

3^/3

Cohorts Resolved

Reached MLB

.103

v0.5 Brier, 79 Outcomes

Numbers Edited

Season complete · All three cohorts graded · Last checkpoint Oct 8 · Model v0.5

The Receipts · 25 Call-Ups

Every Hit, With the Number We Published

The full record

Each of these was locked and published before the player was promoted. The percentage is what we said at the time, not what we wish we had said.

55%YES

JR Ritchie

Debut April 23

Sub-Batch A

50%YES

JJ Wetherholt

Opening Day starter

Sub-Batch A

45%YES

Hagen Smith

Aug 9

Sub-Batch B

45%YES

Angel Genao

Aug 5

Sub-Batch C

44%YES

Lazaro Montes

Sep 1

Sub-Batch C

38%YES

Carson Benge

Opening Day roster

Sub-Batch A

35%YES

Coleman Crow

Spot start April 17

Sub-Batch A

35%YES

Travis Bazzana

Debut April 28

Sub-Batch A

34%YES

Konnor Griffin

Debut April 2

Sub-Batch A

33%YES

Owen Caissie

Sep 9

Sub-Batch B

33%YES

Kaelen Culpepper

Aug 7

Sub-Batch B

33%YES

Max Clark

Aug 1

Sub-Batch C

28%YES

Harry Ford

Jul 17

Sub-Batch C

24%YES

Colt Emerson

Debut May 17

Sub-Batch A

23%YES

Brandon Sproat

Opening Day roster

Sub-Batch A

23%YES

Parker Messick

Won a rotation spot

Sub-Batch A

23%YES

Walker Jenkins

Aug 28

Sub-Batch B

23%YES

Kade Anderson

Aug 22

Sub-Batch B

22%YES

Connelly Early

Opening Day rotation

Sub-Batch A

22%YES

Samuel Basallo

Opening Day catcher

Sub-Batch A

22%YES

Payton Tolle

Debut April 23

Sub-Batch A

22%YES

Joshua Baez

Aug 15

Sub-Batch C

21%YES

Michael Arroyo

Sep 6

Sub-Batch C

14%YES

Cooper Pratt

Debut June 16

Sub-Batch A

13%YES

Kevin McGonigle

Opening Day, 4-for-5

Sub-Batch A

Scroll for all 25 →

Live Layer · Promotion Index · Offseason

The Board Is Quiet, and That Is Correct

Full board + ledger

PIX measures how much promotion pressure is on a prospect right now. It moves only on sourced, logged events, and the 2026 season ended September 27, so there is almost nothing to move it. These are the players still tracked into 2027.

What moves the board this winter: the Rule 5 draft, non-tenders, winter meetings trades, and pre-debut extensions. Each one gets logged with a source when it happens. How PIX works.

Call-Up Probability · Season Complete

Every prediction is graded

The 2026 regular season ended September 27 and the last open window closed October 8. All 107 call-up probabilities published this year have been resolved against binary outcomes and scored. There is nothing open to show here, which is the point: the board is empty because the numbers were settled, not because they were withdrawn.

107

Graded outcomes

Reached MLB

Cohorts resolved

Numbers edited

The October 8 checkpoint graded the largest cohort, 65 predictions locked July 10. It found that the model beat its baseline and that a one-feature model beat the model. Both results are published in full.

Read the October 8 checkpoint The full record

Next cohort locks in spring 2027. The live PIX board continues through the offseason on sourced events only.

Three Checkpoints · Brier vs Baseline

How Each Cohort Actually Scored

Calibration log

Lower is better. The baseline is what you get by ignoring the model and guessing the same number for everyone. Beating it is the minimum bar, and the model cleared it once in three tries.

Sub-Batch AJun 18

14 of 28 reached MLB

Model0.2690

Base0.2500

Lost to the baseline

Sub-Batch BSep 9

5 of 14 reached MLB

Model0.2375

Base0.2296

Lost to the baseline

Sub-Batch COct 8

6 of 65 reached MLB

Model0.0746

Base0.0838

Beat the baseline

The October 8 Finding

Every Call-Up Came From One Level

Triple-A54.5%6/11

Double-A0.0%0/19

High-A0.0%0/23

Single-A0.0%0/8

Complex0.0%0/4

Sixty-five players were scored in July. Six reached the majors and all six were at Triple-A. The other fifty-four, across Double-A, High-A, Single-A and the complex leagues, produced nothing.

A model that knows only whether a player is at Triple-A scores a Brier of 0.0420 on that cohort. The thirteen-feature model scored 0.0746. The simple one is 44 percent better.

Read what that means →

MLB Prospect Call-Up Predictions: How the Model Works

Most prospect coverage focuses on who a player is. Very little answers when they arrive. MKDC Baseball is built around a single question: when will an MLB prospect get called up, and why?

The Call-Up Probability Engine evaluates every tracked prospect across 13 quantifiable features: level proximity, age relative to league, 30-day performance, 40-man roster status, organizational promotion history, positional scarcity, service time considerations, and more. These inputs generate a CUP score (Call-Up Probability within 90 days) and an ECD (Estimated Call-Up Date) via one of three structured pathways.

Why timing is predictable

MLB promotions are not random. Service time mechanics create calendar-driven windows. Organizational behavior is quantifiable. Injuries create immediate openings. The model integrates all three to estimate timing, not just readiness.

What makes this different

Every output is locked and timestamped before outcomes are known. The June 18, 2026 calibration checkpoint is a hard public test. No retroactive adjustments. The model is scored on a Brier score against binary outcomes.

Prospect analysis → How the model works → Baseball metrics explained → How orgs handle prospects →

Four-Stage Pipeline · Model v0.5

How the Engine Works

Full methodology

13 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, Rule 5 flag, and PPI draft-pick incentive. All sourced from public data at evaluation date.

Stage 1–2

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. Locked at run date.

Stage 3

CUP Probability

CUS converts to a Call-Up Probability (CUP), then three multipliers are applied in sequence: org promotion behavior, positional base rate, and service time suppression. CUP is the final number you see in the leaderboard.

Stage 3

ECD Engine

CUP converts to an Estimated Call-Up Date (ECD) through one of three pathways: service time windows (tight range), opportunity or injury (medium range), or performance trajectory (wide range). Each pathway carries a confidence tier.

Stage 4 · v0.5

Every Number,Graded.

Every prediction is graded

MLB Prospect Call-Up Predictions: How the Model Works

Every Number,
Graded.