Another part of the continued playoff-chances analysis, this week I looked into how much travel mileage impacts a teams postseason odds.
The results were pretty surprising, as many sports news outlets often parrot it being a very bad thing (at least on a game-by-game basis). This is often a factor considered in serious sports betting and fantasy football communities.
The concern mostly stems from timezone changes, especially for west coast teams traveling east (where 12PM ET games feel like 10AM PT). This analysis does not particularly look at timezone changes, more so flying mileage.
Data
The data was a little easier to get than my last playoff-chance analysis. The first thing that I needed to do was establish some assumptions:
Travel distance is measured between both team's nearest airports. I did this to avoid the subtle, yet often-enough-to-be-annoying changes in stadiums (example: the Cowboys moving from Irving, TX to Arlington, TX or weird temporary years like when the Vikings played at the University of Minnesota after their roof collapsed).
I established the Baltimore-200 Rule. This rule is a rough delineation between when a plane is used and when it isn't. Baltimore was used based on this article that highlighted how much Baltimore uses alternative means of travel given their proximity to so many other teams. The "200" comes from Baltimore's furthest-closest opponent being the Meadowlands, NJ, which is 195 miles away.
The airport assumption also applies to international games. In some cases with international games (really just London), the stadium will change. Again, I wasn't going to deal with the hassle of minute latitude-longitude changes between stadiums.
To get airport latitude-longitude of US airports, I used this data source. I then needed to create a table that had every unique NFL city since the 1966 season and its respective latitude-longitude coordinates of its airport (yes I know some airports even move, but that wasn't taken into account because, like stadiums, it would be negligible). This was handy as some teams have shared cities over the years and this allowed me to just key off of it.
Year | Team | City | Long | Lat |
|---|---|---|---|---|
1966 | NEP | Boston | -71.00520325 | 42.36429977 |
1966 | BUF | Buffalo | -78.73220062 | 42.94049835 |
1966 | TEN | Houston | -95.27890015 | 29.64539909 |
1966 | MIA | Miami | -80.29060364 | 25.79319954 |
1966 | NYJ | New York | -73.87259674 | 40.77719879 |
1966 | DEN | Denver | -104.6729965 | 39.86169815 |
1966 | KCC | Kansas City | -94.713898 | 39.2976 |
1966 | LVR | Oakland | -122.221001 | 37.721298 |
1966 | LAC | San Diego | -117.1900024 | 32.73360062 |
Pain In The A** Alert: To consider international locations, I also hard-coded the latitude-longitude locations of the international city airports (there were few and repeatable enough to hard-code, I wouldn't recommend it if they weren't). I incorporated a flagging system that duplicated the game-by-game data and switched the home team to being the visitor because they too would be traveling. The flagging system was not elegant, to say the least. I essentially needed to flag the international game based on the venue, copy the game as another row, flag THAT row with a different flag so the distance could be calculated using the home teams location later.
After that, it was pretty straightforward in aggregating the teams travel distance per season. There was a minor issue with AFL teams pre-merger, but that was resolved quickly.
The raw data alone provided some interesting information, including the top distances traveled ever in the NFL.
Year | Team | Distance Traveled |
|---|---|---|
2018 | Chargers | 21,910 miles |
2013 | 49ers | 21,278 miles |
2021 | 49ers | 19,181 miles |
1986 | 49ers | 18,926 miles |
2012 | Seattle | 18,777 miles |
1989 | Rams | 18,168 miles |
2014 | Raiders | 18,060 miles |
2016 | Rams | 18,005 miles |
1984 | 49ers | 17,961 miles |
Unsurprising, these are all west coast teams. And in a lot of cases, many of these (post-2007) had an international trip included. This will become an important takeaway discussed in the results.
To see the full code and data - be sure to check out my Github.
Results
Well, the results were initially shocking, and had me racking my brain for a while in trying to explain it. To start, I ran a simple histogram to see distribution.
The plot shows a semi-skewed normal distribution with the average hovering around 7500-8500 miles. The long tail to the right can also partly be explained by the introduction of international games and how relatively few there have been (yet significant addition to travel distance).
Now the weird part. When including playoff percent with the same bins, a trend appears that seems to suggest more traveling is a good thing (or betters a teams playoff chances).
A pretty clear increasing trend shows that with greater miles traveled, playoff percent increases.
My first reaction was to remove the outliers based on z-score and a threshold of 3 standard deviations. In many texts, this is a good measure for outlier. After doing that, the following charts were generated.
It helped, but not much. There's still something odd here - or is there? Going through the raw data, the trend is indeed accurate. In fact of the top 25 distances ever traveled, 19 of the teams made the playoffs. WTF!
After a long period of stewing I can only make 2 conclusions as to why this is, and neither is anything like "if they are on planes longer, they can overthink or overplan less" or something silly like that.
Good teams are generally immune to the negatives of long travel
It makes sense as good teams find ways to make it happen. With the historical prestige of teams like the 49ers, traveling further than most teams just simply didn't impact their performance.
It might be a timing thing
Could this be a timing thing? The international series started in 2007 and around that time, many west coast teams saw a strong, consistent surge in performance. Since 2007, we've seen Seattle in 2 Super Bowls, the Rams in 2, the 49ers in 3, and the Cardinals in 1. Throw in teams in the middle of the country and you add Denver in 2 and the Chiefs in 4. The moral of the story being related to the first point - west coast teams have proven to be of championship caliber and are obviously traveling more than most.
So, going forward, I don't think I'd recommend teams increase their travel distances for fun. At the same time, I'll now question the sports media world when they claim the detrimental impact of travel distance.
Thanks for reading!
More to come in the sports world... and Love Is Blind?
-Harrix
Comentários