046: Questioning game duration data

A data project that tracks the duration of 2023 college football games.

👋 I’ve been working on a project to surface the length or duration of college football games. This data can be used to analyze recent rule changes or measure pace of play.

You can check it out here:

wHy Do YoU wAnT lEsS fOoTbAlL?

If you think tracking the length of football games is about wanting less football, you’re missing the point.

The flow of the game is undesirable, but the intention of the rule changes was not to expedite the game¹.

The goal of tracking the data is to ask better questions.

Why do games take so long?

The games are long, but I’m skeptical we know all the reasons why?

Sure, commercials and TV have a ton to do with it. But how much?

What about the pace of the two teams? Injury timeouts?

Official reviews? The weather?

How are teams using the clock?

Given the rule changes mean fewer plays per game, it also means fewer possessions or drives.

So, what does that mean for game totals?

And does that mean more blowouts or more close games?

How does each team handle a drive and when do drives occur?

Is the first quarter faster than the other quarters?

What does the data say so far?

Through 94 games this season, the average plays per game for one team is 65.8 and the average game duration is three hours and 24 minutes.

This is about three fewer plays than a season ago, and three fewer minutes of games duration.

Plays per game are defined by offensive plays. This means adding rushing attempts and passing attempts by team. It excludes punts, field goals, kickoffs.

The game duration is adjusted at times for weather delays. For example, the game between Arizona State and Southern Utah lasted over six hours. The caveat is there was a two-hour and 50 minute weather delay.

Drives

The average number of drives per game for one team so far is 11.9. When broken out by half, this averages out to 6.2 in the first half and 5.7 in the second half. These numbers include “garbage time” drives².

# drives by team by game and half
                  value
avg_drives_game   11.9
avg_1st_half       6.2  
avg_2nd_half       5.7

Chip Kelly was not happy about only having four drives in the first half. Both UCLA and Coastal Carolina had seven drives in the second half, as the Bruins won 27-13.

Here is the distribution of number of drives for each half:

Quarters

The data for duration by quarter is difficult to track without reliable play-by-play data. It can be done manually, and I tried that for the North Carolina and South Carolina game.

# duration in minutes 
# by quarter
              minutes
1st_quarter   ~35
2nd_quarter   ~45   
3rd_quarter.  ~37
4th_quarter   ~46

The first and third quarters were both under 40 minutes, while second and fourth quarters were both 45 minutes or longer. Add in a 20-minute plus half-time and a few minutes between each quarter, and you arrive at 193 minutes or a three hour and thirteen minute game.

There was a bizarre sequence in this game too. North Carolina elected to not call a timeout at the end of the first half with a minute and nine seconds remaining, and kneeled out the clock to forfeit a possession. Half-time was longer than 20 minutes because the chain crew was eating hot dogs late, and then South Carolina recovered an onside kick.

Here is 36 seconds of the chain crew being late . . .

A total of 28 minutes of real time elapsed between South Carolina not converting third down at the end of the first half and recovering the on-side kick to start the second half. It was only about one minute and ten seconds of actual game time.

---

My plan is to continue to track this data throughout the entire 2023 season. I’m working on automating the data updates, adding a new page to see duration by team and drives by half, and still sorting out how to account for total plays and weather delays.

If you have any recommendations or feedback on how to make site more useful, please let me know.

1. Again, this is what Alberto Riveron said at the ACC media days. Not to expedite the game, it’s only for player safety. Riveron’s claim to make reviews faster feels ambitious so far.

2. Garbage time drives are drives that occur during a blowout, or when the team simply takes a knee to end the half or game. This is sort of subjective to filter out of the data, so leaving it as is for now.