Do Pitchers Lose Their Command?

The Question

In our earlier analysis, we found that the correlation between pitch count and miss distance is r = 0.007 — essentially zero. We concluded that pitchers don't measurably lose their command.

A reader posed the right follow-up: what about variance? The average miss distance might stay flat, but does the frequency of severely mislocated pitches increase? Is the broadcaster wrong about the average pitch but right about the worst ones?

We tested this with the full 2025 MLB season: 729,827 pitches, 4,892 true starter outings of 30+ pitches, 185 pitchers with 10 or more starts.

The Average Is Flat. The Distribution Is Not.

First, the average. Plate-location scatter — the combined standard deviation of where pitches cross the plate — is essentially constant across a starter's outing:

The mean doesn't move. But the distribution of outing-level scatter change tells a different story:

Of 4,892 starter outings in 2025, 49.7% show essentially no change (scatter moves less than 10% either direction). But the tail is asymmetric: 14.0% of starts see scatter increase by more than 20%, versus only 5.2% that tighten by the same amount. A 2.7-to-1 ratio.

This isn't a pitch-mix artifact: after demeaning within pitch type, the rates are still 12.9% blow-up vs 5.9% tighten.

What It Looks Like

Spray charts make the pattern visible. Here are four recent outings from the 2026 season — three starts and one long relief appearance — showing how scatter changes within an outing:

Roddery Muñoz's scatter nearly doubles (+55%) during a 47-pitch long relief appearance on March 29 — his late pitches land in the dirt, above the batter's head, and two feet off the plate. (Muñoz entered in the 3rd inning as a reliever, not a starter — but the scatter pattern is the same.) Taijuan Walker's start (+48%) includes pitches that sailed 6 feet above the ground. Casey Mize (+42%) shows the same pattern across 93 pitches. But Roki Sasaki tells the opposite story — his scatter drops 25%, getting more precise as his start progresses. View these game reports →

The Outing Length Puzzle

If scatter blow-ups were caused by fatigue, you'd expect longer outings to have the highest blow-up rates. The opposite is true:

Short starts (30-59 pitches) have a 23.4% blow-up rate — nearly double the rate for long starts (90+ pitches, 12.2%). The most likely explanation is selection bias: pitchers pulled early are often pulled because they're wild, not wild because they're tired. The starters who go deep tend to maintain their scatter.

This complicates the fatigue narrative. The data doesn't say "pitchers lose command because they throw too many pitches." It says "a minority of outings produce scatter spikes, and those outings tend to end early."

The Pitcher Profiles

The most actionable finding: scatter blow-ups are heterogeneous. Some pitchers are chronically blow-up-prone; others essentially never blow up.

Most blow-up-prone (2025, starter-only, 10+ starts)

Garrett Crochet is the extreme case: his plate-location scatter increases by 31.5% on average from early to late in the outing, across 33 starts. Carmen Mlodzinski blows up in 58% of his starts — the highest rate of anyone with 10+ starts.

Most consistent (2025, starter-only, 10+ starts)

Ben Brown gets tighter as the game progresses — 12% more concentrated in the last third vs the first, with zero blow-up outings in 15 starts. Pivetta maintained his scatter across 33 starts with 0% blow-up rate.

What We Can and Can't Say

What the data shows: Average plate-location scatter is flat across a starter's outing (confirming r = 0.007). But the distribution is asymmetric — 14% of starts produce a significant scatter spike, vs 5% that tighten. The effect is heterogeneous: some pitchers are chronically blow-up-prone, others never blow up.

What this metric is: Plate-location dispersion — where the ball crosses home plate relative to the pitcher's typical location for that pitch type. This is a proxy for command, not command itself. We don't have catcher target data, so we can't measure how far pitches land from their intended location. Research suggests release-point variability is the primary driver of location (Kusafuka et al. 2020).

What we can't say:

• We can't confirm this is fatigue-driven. The outing-length data (short starts blow up more) suggests selection bias, not gradual degradation.
• We can't yet say whether "blow-up-prone" is a stable pitcher trait across seasons. Early 2026 data is suggestive (Mlodzinski: +20.7% in 2025, +18.8% in 2026) but based on only 2 starts per pitcher.
• We can't rule out that some scatter changes reflect intentional pitch-mix shifts (though the pitch-type-adjusted check suggests this isn't the main driver).

The revised conclusion: The broadcaster saying "he's losing his command" is wrong about the average pitcher on the average night. But about 1 in 7 starts, the scatter does spike meaningfully — and knowing which pitchers are prone to it is genuinely useful information.