Inside the Numbers Projection vs. Engine

Ray Mileur
Feb 21, 2026By Ray Mileur

Projection vs. Engine: What the Simulations Really Reveal
When belief in development meets current reality.

Before Opening Day, two very different verdicts emerged on the 2026 St. Louis Cardinals.

The first — built on 10,000 projection-based simulations — suggested a hinge-year club hovering around .500. A team dependent on growth. A roster with swing variables. Median outcome: 81–81.

The second — a full 162-game simulation in MLB The Show 25 — was less forgiving. Final record: 69–93. Fifth place. –131 run differential.

The gap between those outcomes is not just numerical.

It is philosophical.

The projection model allowed for development. It assumed that some young players improve, some stagnate, and a few surprise. It modeled variance. It built in aging curves and breakout probabilities.

The game engine did not care about probability. It cared about ratings.

Entering the season rated 29th overall and 29th in pitching, the Cardinals were treated as a bottom-tier club — and they played like one. The engine does not forecast potential. It measures present ability.

The projection said, “What if growth happens?”
The engine said, “Show me.”

And that is the tension facing this organization in 2026.

If JJ Wetherholt is a .250 rookie with moderate power, if Jordan Walker and Nolan Gorman remain inconsistent, if the rotation performs to current metrics rather than future promise — the 69-win result is not outrageous.

But if two or three young pieces make legitimate jumps, if the rotation stabilizes, if development accelerates instead of stalls — the 81-win projection becomes realistic.

What these simulations reveal is not certainty.

They reveal the margin.

The 2026 Cardinals are a development-dependent team. Their range of outcomes is wide because their foundation is still forming.

One model saw potential.

One engine saw present limitation.

The truth, as always, will be found somewhere between belief and evidence.

And over 162 real games, the players — not the math or the machine — will settle the debate