We logged every output publicly on March 20 before a single game was played. One already resolved. Twenty-three still open. Here's where the model stands on Opening Day.
We ran 27 MLB prospects through the model on March 20. Every output got logged publicly before a single game was played. Today that window opens.
Carson Benge is in the Mets' Opening Day lineup as the starting right fielder. The model had him at 38% CUP, the highest score in the batch, with an ECD of April 10. He resolved six days after run date. The features pointing toward an early call were real. A genuine vacancy in right field after Soto moved to left, a strong spring, PPI eligibility with the deadline approaching, and an org with reason to act early. Tauchman tearing his meniscus removed the last uncertainty.
One resolved outcome. Twenty-two still open.
Wetherholt, McLean, Painter, and DeLauter came in as confirmed before scoring, so they resolve as expected. Kevin McGonigle making Detroit's Opening Day roster is the biggest prospect news of the last 24 hours. He was not in this batch, but the Tigers skipping him straight from Double-A is the kind of org behavior worth understanding. Detroit's promotion history would likely score well in the model. Worth factoring into Sub-Batch B.
MKDC Baseball tracks call-up timing. Not talent. Not fantasy value. The model uses 12 quantifiable features to estimate whether a player makes an MLB appearance within 90 days of the run date. Every number on the leaderboard was locked on March 20 and stays locked. Nothing gets adjusted retroactively.
The features are things you can source from public data at evaluation date: level proximity, age relative to league average, recent performance, 40-man status, org promotion history, position depth, options remaining, spring signal, PPI eligibility. No scouting reports. No gut reads.
This is the model's first real season. v0.4 is built on calibrated estimates rather than trained outcomes. What this season produces, every resolved prediction between now and June 18, becomes the data that informs whatever comes next. The goal for 2026 is finding out where the model is wrong as much as where it is right.
Fourteen days away. Three players flagged for PPI downgrade if they are not on an active roster by end of day: Emerson (SEA), Clark (DET), White (MIA). The calibration log gets updated the same day outcomes resolve. Hits and misses both.
June 18 is when the first real accounting happens. The outcomes between now and then are what the model learns from.
Full leaderboard at mkdcbaseball.com. Calibration log at mkdcbaseball.com/calibration. Methodology at mkdcbaseball.com/methodology. Follow @MKDCBaseball on X for same-day updates when outcomes resolve.
Back on April 9.