NFL Indoor vs. Outdoor Performances

Especially in the winter time we hear about NFL teams performing differently when a team with an indoor stadium has to play outdoors, but do teams actually perform worse in these situation? What about in the rain? Or just outside in general? Do teams pass the ball better inside? Do teams run more in the rain?

This article answers all of those questions.

For the analysis, I used NFL play-by-play data from the 2016 – 2019 regular and postseason

---

First, I looked at completion percent over expectation (CPOE) (this is a measure how how good a throw is, higher CPOE means a pass was completed when it had a low completion probability) by air yards for three different game environmental situations: indoor games, outdoor games, and outdoor games where it was raining.

We see that indoor and outdoor games really follow the same track, but when it’s raining throws are completed at a lower level. It is, of course, harder to complete passes in the rain.

What about running plays? Below I plotted expected points added (EPA) and rush attempts for the same three situations.

Running is a different story. Here we see all three really follow the same path. It appears running the ball is not impacted by the different situations.

---

Does temperature make a difference? Not really. Shown below, EPA per play does not change as the temperature increases. Games played at 0 degrees have a very similar distribution as games played at over 90 degrees.

But what about passing specifically? Below is CPOE by temperature.

There is a difference here! At temperatures below 40 degrees CPOE is below 0 on average. At temperatures above 40 we have CPOE above 0. Clearly temperature impacts the quarterback and receivers.

We see below too that passing the ball is much more efficient than running the ball as temperatures increase. The most efficient passing temperature is around 60 degrees, whereas running efficiency is best at around 40.

Now, correlation does not mean causation, but it is interesting to see running plays become less efficient as the temperature increases while passing plays do the opposite – become more efficient as the temperature increases.

---

Finally, I looked at how teams performed when they were playing in situations other than their home field:
How do indoor teams perform outdoors?
How do outdoor teams perform indoors?
How do teams perform in their same home situation?

First, I looked at passing.

The yellow line shows teams playing in the same indoor/outdoor situation are very average, which would make sense.

The green line shows outdoor teams playing indoors play above average with passes less than 5 yards, below average with passes between 5 and 30 yards, then are back above average for long passes over 30 yards. Maybe quarterbacks are over throwing targets at medium depth, but are able to really throw the ball deep?

The red line shows that indoor teams playing outdoors are above average at passes less than 5 yards, have a small dip below average, then are above average with targets deeper than 15 yards.

A similar analysis was done for running plays.

Not a lot of variances between these. A slight advantage in a high number of rush attempts for indoor teams playing outdoors (red line), but really it’s consistent across the board. There doesn’t seem to be an advantage for running the ball depending on where a team is playing.

The passing result surprised me. Often we hear about outdoor teams being able to throw the ball deeper than normal in a dome, which we do see the the data support, but we never hear about indoor teams completing passes better when playing outdoors, which the data also supports.

---

So does playing in a different situation help or hurt a team? Does temperature or weather impact how teams play? The data suggests we do see some performance differences when passing the ball, but overall and with running the ball the differences are minimal. Teams do play different when playing in a different stadium environment than their home, but noticeable changes are seen in just fairly specific situations.

Data: nflfastR