Baseball's conventional wisdom is full of testable claims. We tested two of the most common — and neither survived contact with the data.
Myth 1: "He's Losing His Command"
Every baseball broadcast has this moment. The pitcher walks someone in the sixth inning, and the announcer says: "He's starting to lose his command out there." It sounds right. It feels right. Pitchers get tired. Their mechanics break down. The ball starts missing its spot.
Except the data says this barely happens.
We computed a "miss distance" for every pitch thrown by a starting pitcher in 2025: how far the pitch landed from where that pitcher typically locates that pitch type in that count. Then we checked whether miss distance grows as the game progresses.
| Pitch count | Mean miss distance |
|---|---|
| 1 – 25 | 11.17 inches |
| 26 – 50 | 11.16 inches |
| 51 – 75 | 11.21 inches |
| 76 – 100 | 11.34 inches |
The numbers barely move. Here are the formal statistics:
| Metric | Value |
|---|---|
| Correlation (pitch count vs miss distance) | r = 0.007 |
| Statistical significance | p < 0.00001 |
| Slope | 0.018 in per 10 pitches |
| Natural variation (std dev) | 6.46 inches |
Yes, the correlation is statistically significant — with 729,827 pitches, almost everything is. But r = 0.007 means pitch count explains 0.005% of the variance in miss distance. A pitcher's 100th pitch of the game misses his target by about a fifth of an inch more than his first pitch, against a natural variation of 6.46 inches. The signal is invisible inside the noise.
So why do pitchers perform worse late in games?
The "losing his command" narrative is almost entirely an attribution error. When pitchers perform worse late in games, it's primarily because batters see them better the third time through the lineup — the well-documented times-through-the-order effect. It's the batter adjusting, not the pitcher deteriorating.
The distinction matters. If the pitcher is losing his command, you should pull him when his mechanics degrade. If the batter is adjusting, you should pull him based on times through the order regardless of how "sharp" he looks. The data strongly supports the second interpretation.
Myth 2: "Plate Discipline Predicts Future Hitting"
This one hurt to kill because the logic is so appealing. A batter who makes good swing/take decisions should perform better in the future, right? A disciplined hitter in a slump is a "buy" signal for fantasy. An undisciplined hitter on a hot streak is a "sell."
We tested this systematically across 522 qualified batters from the 2025 season.
Step 1: Build a discipline model
First, we built a swing prediction model — given the pitch location, velocity, movement, and count, how likely is the average batter to swing? This model achieved 0.81 AUC, meaning it reliably captures what a "correct" swing decision looks like.
A batter's discipline score is the gap between the model's expected swing rate and their actual swing rate, computed over a rolling 7-game window. Positive scores mean the batter is more selective than the model expects; negative scores mean they're chasing more.
Step 2: Does discipline predict forward performance?
We correlated each batter's 7-game rolling discipline score with their wOBA over the next 7 games.
| Metric | Value |
|---|---|
| Correlation (discipline vs future wOBA) | r = -0.019 |
| Batters with positive relationship | 49.4% |
| Qualified batters analyzed | 522 |
The correlation is r = -0.019. Exactly 49.4% of batters showed a positive relationship between discipline and future performance — indistinguishable from flipping a coin.
What's going on?
Discipline is real. Discipline is measurable. But discipline does not predict future hitting. Why?
The most likely explanation is that discipline is already priced into the current hitting results. A disciplined batter's selectivity shows up immediately in their walk rate, their in-zone swing rate, and their current-period wOBA. There's no residual predictive power left over for the future period. The market is efficient.
This doesn't mean plate discipline is worthless — it clearly helps batters reach base. It means you can't use it as a leading indicator to predict breakouts or identify buy-low candidates. The fantasy baseball dream of "discipline predicts slump recovery" is not supported by the data.
Why We Publish Negative Results
Both of these findings are negative results. Neither confirms a useful product hypothesis. Neither makes for a headline that says "We Found Something Amazing."
We publish them because negative results are the most underreported findings in sports analytics. Confirmation bias means the analyses that confirm conventional wisdom get published and shared. The ones that find nothing get filed away quietly. That's a problem, because the null findings are often more useful than the positive ones — they tell you where not to look, which is just as valuable as telling you where to look.
Pitchers don't measurably lose their command. Plate discipline doesn't predict future hitting. These aren't opinions — they're findings that emerged from data and survived independent review.
Methodology
Data source: 2025 MLB Statcast data (729,827 pitches). Accessed via pybaseball.
Command analysis: Miss distance = Euclidean deviation from pitcher's mean location for that pitch type in that count. Computed for all starting pitcher pitches. Rolling 15-pitch window. The forward-looking target uses a properly shifted 3 at-bat window to avoid contamination.
Correlation: Pearson r between pitch number within game and miss distance. The p < 0.00001 significance is expected with n = 729,827 and does not imply practical significance.
Discipline analysis: Swing prediction model (logistic regression, 0.81 AUC) trained on pitch location, velocity, movement, and count. 7-game rolling discipline scores for 522 qualified batters (minimum 200 PA), correlated with next 7-game wOBA.
Qualified batters: Minimum 200 plate appearances in 2025. 522 batters met this threshold.
If you find an error, tell us — we'd rather be corrected than wrong.