Two analyses across 6 million pitches and 9 MLB seasons reveal that travel fatigue is almost entirely a reliever problem — starters are effectively immune.
Before looking at performance, the raw scheduling data tells the story. Starters get twice the rest, travel less frequently, and face far fewer compressed turnarounds.
Version 2 split the same 6M pitches by role. Every travel-related correlation with pitch quality is 2–4x stronger for relievers. Starters barely register.
All correlations significant at p < 0.001. Positive r = more rest/distance correlates with better pitch quality.
Version 2 introduced elapsed hours — the actual time between game starts, not just calendar days. Relievers show a clear gradient: tighter turnarounds produce worse pitches. Starters are flat.
The 18–30 hour window — a next-day appearance with a tight turnaround — is the worst spot for relievers. Starters show effectively zero variation across rest levels.
Is it the miles that hurt, or the clock? The distance × time interaction matrix shows that for relievers, time pressure is 3–4x more important than travel distance.
A rested reliever pitches well regardless of how far they traveled. A rushed reliever pitches poorly regardless of whether it was a short hop or cross-country flight. The clock dominates the map.
Version 2 introduced "travel rate" — miles traveled divided by hours since last game. This captures the rushedness of travel in a single number. For relievers, it produces the cleanest monotonic degradation of any feature tested.
The swing from easy (<5 mi/hr) to rushed (30–50 mi/hr) is ~0.004 xRV per pitch — about 4 hundredths of a run. The extreme bucket (50+ mi/hr) levels off, likely because those situations are so rare that managers avoid using key relievers.
Before splitting by role, version 1 found an interesting pattern: the biggest jump is from "no travel" to "any travel." Beyond that, distance barely matters.
"Same city" includes homestands and short hops. The positive residual for travel games is likely selection bias (teams send rested pitchers for road trips). The key finding: 500 mi and 2000 mi look identical.
2-timezone crossings are the peak effect. 3-timezone crossings don't increase it further — potentially because coast-to-coast trips involve off-days built into the schedule.