Travel, Altitude, and the Factors People Overrate

Somewhere in every prop research community there is a person who knows exactly how many miles a team flew this week. They will tell you about the two time zone shift, the early local start, the thin air at the destination, and the fact that this is the fourth city in nine days. The presentation is compelling because it is specific, verifiable, and almost nobody else bothers to look it up.

It is also, for most player props, close to worthless. Not entirely worthless. This article is not an argument that environment never matters. It is an argument about magnitude, and about a failure mode that appears whenever a researcher collects several small factors that all point the same way and mistakes the sum for knowledge.

Altitude is the strongest member of this group, and it is still small

Altitude earns more respect than the rest of the category because it has an unambiguous physical mechanism rather than a psychological story. Air at elevation is less dense. That is a fact about the atmosphere, not about how tired someone feels, and it acts on two separate channels at once.

The first channel is ball flight. Lower air density means less drag, so a struck or kicked ball carries further and breaking pitches move less. This is a mechanical consequence, and it applies identically to every player on the field regardless of conditioning. The second channel is human physiology. Lower partial pressure of oxygen means less oxygen delivered per breath, and the body's responses to that are well documented in the literature on the effects of high altitude on humans. Endurance work gets harder, recovery between efforts slows, and the effect is largest for athletes who arrived recently and have not acclimatized.

So altitude is real. The question a researcher actually has to answer is different: is it large enough, relative to everything else moving on a given night, to change a projection by an amount worth acting on? For a single player prop, the honest answer is usually that it is a minor term. The ball flight channel shows up mostly in outcome distributions rather than in individual volume stats, and the physiological channel is competing with variables that swing far harder, chief among them how much the player actually plays.

Travel distance and time zones are widely cited and weakly supported

Flight distance and time zone crossings occupy a strange place in sports analysis. They are cited constantly, they sound scientific, and at the level of an individual player's counting stat they are difficult to detect at all.

Part of the problem is that the effect, if present, is small relative to the noise in a single game. A player's line to line variation in a volume stat is enormous; a factor that might shift a true rate by a percent or two is buried under that variation and would take a very large sample to isolate. Part of the problem is that travel is confounded with everything else. Teams that fly the furthest are also the teams playing the most compressed schedules, facing unfamiliar opponents, and dealing with rest patterns that would explain the same observation. Attributing the result to the flight rather than the schedule is a choice, not a finding.

And part of the problem is that professional organizations spend real money specifically to neutralize this. Charter flights, arrival timing, sleep protocols, and travel staff exist because the effect is understood and being actively managed. A factor that a well funded opponent is deliberately mitigating is a poor foundation for an edge claim.

The version of this that does hold up arrives through the schedule rather than the map. Games in short succession, minutes played recently, and the pattern of rest are observable, they are directly connected to how a coach will deploy a player, and they are already covered in rest and schedule spots. If a road trip matters tonight, it usually matters because it produced a rest situation that changes playing time, and playing time is a channel you can research directly instead of inferring it from air miles.

Why this whole category is so seductive

It is worth being explicit about why these factors keep reappearing despite their weak support, because understanding the appeal is what makes it resistible.

• They feel like hidden knowledge. Everyone can see a player's recent averages. Fewer people looked up the elevation of the venue or counted the time zones crossed, and the scarcity of the information is mistaken for the value of it.
• They are easy to obtain. Elevation, flight distance, and local start time are all a lookup away, while the genuinely important inputs, such as tonight's rotation or a role change, may be unresolved until shortly before the event.
• They always support a story. Any of these factors can be narrated toward the over or the under depending on which conclusion you already prefer. Thin air helps the ball carry; thin air tires the legs. Both sentences are true and they point in opposite directions.
• They are unfalsifiable in a single game. If the player goes over, the factor did not matter that night. If he goes under, the factor is confirmed. Nothing about one result can dislodge the belief, which is the defining property of an explanation that is not doing any work.

That third point deserves emphasis. A factor that can be argued in either direction is not an input, it is a rhetorical device. Before you allow an environmental adjustment into a projection, write down which direction it moves the number and why, ideally before you know which side you find interesting. If you cannot commit to the direction in advance, the factor is decoration.

The real damage is stacking, not any single factor

One small environmental adjustment applied honestly is nearly harmless. It moves a projection slightly, and the movement is dominated by the uncertainty already present. The failure mode is compounding: altitude, plus travel, plus weather, plus an official's tendencies, plus a stadium note, all applied to the same projection, all pointing the same way.

Two things go wrong at once. First, the adjustments are not independent. Weather, venue, and altitude are partly the same fact described three times, and applying them separately counts the same evidence repeatedly. Second, and more corrosive, the researcher's confidence grows with the number of factors considered even though the actual information did not. Five weak reasons feel much stronger than one weak reason. They are not.

Notice that in the example every adjustment was small and none was absurd. That is precisely the danger. Obviously bad inputs get caught. Plausible small ones accumulate quietly, and because they were selected after the researcher had a leaning, they are correlated with the leaning rather than with reality.

A rough hierarchy of what actually moves a player prop

Ordering matters more than any individual judgment here. If you spend your research time in order of expected impact, the environmental tier rarely gets reached, and that is the correct outcome.

1. Availability. Whether the player is in the event at all dominates everything else, since a missing player produces either a zero or a voided leg depending on the platform.
2. Opportunity and role. Minutes, plate appearances, snaps, possessions, batting order slot, jungle role, service games. Volume multiplies into every counting stat, and a role change resets the meaning of a player's whole sample.
3. Skill rates and their uncertainty. The player's own underlying rate, honestly shrunk toward a baseline in proportion to how thin the sample is.
4. Direct matchup structure. Opponent quality specifically as it applies to this stat, plus anything structural such as the expected pace or length of the event.
5. Game script. Whether the projected flow of the event gives this player more or fewer chances than a neutral script would.
6. Environment. Altitude, weather, venue quirks, officiating tendencies, and travel. Real, occasionally worth a small nudge, and never a reason to override anything above it.

Weather deserves one carve out, which is that it graduates tiers in specific outdoor situations. Strong wind or heavy precipitation is not a small nudge to a passing volume prop or a home run market; it can change the fundamental shape of the event. Outdoor conditions covers where that line sits. The general principle survives: a factor belongs in the environment tier until its mechanism is large and specific enough to promote it, and the promotion has to be argued rather than assumed.

Practical discipline: one adjustment, capped, and never a rescue

Three rules keep this category useful without letting it take over.

• Allow at most one environmental adjustment per projection, and pick the one with the clearest physical mechanism. If altitude and travel and officiating all seem relevant, they are mostly relevant to your confidence, not to your number.
• Cap its size. An environmental factor should move a projection by a small fraction of its own uncertainty. If an adjustment is large enough to flip your conclusion, it is being asked to do work it cannot support.
• Never let it override an opportunity input. If the projection says pass on the strength of minutes or lineup position, an elevation note does not rescue it. Rescuing a pass is the single clearest sign that the adjustment is serving the conclusion rather than the analysis.

There is a useful cross check here too. When your read and another careful source's read differ, the discipline described in why models disagree applies: find the input responsible rather than picking a winner. If the only thing separating you from a reasonable alternative view is a stack of environmental nudges, you have just diagnosed your own analysis, and the honest response is to drop back to the number you had before the nudges.

How Slateline treats environmental inputs

Our engines model environment where the mechanism is structural rather than atmospheric. Park geometry and weather feed the baseball simulation because they change the physics of a batted ball. Schedule spots feed the basketball engines because they change availability and deployment. Court surface feeds tennis because serve and return rates genuinely differ by surface. In each case the factor is modeled as a measured input inside the simulation, not applied afterward as a manual nudge to a finished number.

What we deliberately do not do is author small coefficients for effects we cannot measure. A travel penalty invented to make a projection feel more sophisticated adds no information and hides behind the credibility of the parts that were measured. When a factor is unverified, the honest treatment is a wider distribution or a caution flag, not a confident tilt, and both of those show up on the signal board rather than being buried.

The habit worth keeping is simple. Rank your inputs by how much they can move the outcome, spend your attention at the top of that list, and treat everything at the bottom as a tiebreaker at most. Whether that discipline has actually paid off is a question for the graded record, which is public in the Model Room, misses included.