Baseball game length: A visual analysis

Every post has three important phases -- the idea, research to support the idea and actually writing it. As a former high school and college debater back in the days before electricity, the second and third parts come fairly easily, and it's the idea that I consider to be the coin of the realm. As such, I take them where I find them.

I saw a tweet from Patrick Dubuque, an editor for the SB Nation Mariners site Lookout Landing. He was asking some general questions about game length, and since no one asked me I sent him my go-to chart on game length:

Baseball-Reference has the game length data back to around 1920, and I refined it by limiting it to nine-inning games or less. Since no one tons of people find this interesting whenever I post it on Twitter, I threw it up on the Beyond the Box Score Facebook page (like us today -- please!), which elicited a response from BtBS alumnus Bill Petti to add runs scored. This took five minutes (four for the database to open, one to make the chart):

Longer games and fewer runs scored, a sure-fire recipe to engage the next generation of baseball fan. A cursory look reveals these data don't tell the entire story as the chart shows only runs scored by one team, not the total runs scored in the game by both teams -- the fact a team scored one run in no way precludes the other team from scoring twenty, so this wasn't enough.

I currently have data from 1920-2014 and added fields for total plate appearances and innings pitched for games, not just for individual teams. The Tableau data viz shown below has six tabs, which I will explain in great detail:

Important note -- the underlying data is a spreadsheet with 350,000 rows and 60 or so columns. If you choose to download the viz, it might take a while.

The first tab, Game Length, is traditional game length data, with an option to filter by total number of innings for both teams. I have filtered it to games between 17-18 innings, the typical game. Use the slider to change the number of innings, but note that nine-inning games represent around ninety percent of games. Consider these factors for game length of a typical nine-inning game:

Component	Number	Time
Total pitches (25 second pace)	300	2:05
Inning breaks (2 minutes)	18	0:36
Other	??	0:21

I'm assuming 150 pitches per team per game and used twenty-five seconds between pitches, which is actually a little high as the pace was twenty-three seconds in 2014, but I like round numbers -- using the 2014 value would reduce pitch time by ten minutes and increase other time by ten minutes. Having said that, clicking the pace link on the data viz shows that the time between pitches has increased by 1.5 seconds since 2007, which amounts to almost eight minutes a game. It's pretty clear in the charts.

Inning breaks are a standard two minutes, although I've never taken out a stop watch to actually time them. The Other category includes balls in play, mid-inning pitching changes, trips to the mound, replay reviews and things along those lines. Putting it at "only" twenty-one minutes seems generous.

The second Tableau tab, Plate Appearances, begins to get to the heart of the matter. Teams can play nine-inning games, but a 1-0 slugfest will be quicker than a 14-12 pitching duel, and the data viz shows how game length can be affected by the total number of plate appearances. I set it between 70 and 80 total plate appearances, but it can be moved around as desired.

The Game Length by Total PA tab shows the very direct relationship between total plate appearances and game length. The Total PA tab shows the distribution of games and reveals the majority of games have between 67-83 total plate appearances for both teams. Pitches/PA shows the increased patience batters are demonstrating in their plate appearances -- seeing more pitches will result in longer games. The last tab is Runs Scored, a fun one that shows the change in game length for games with a given score -- I've set it at 1-0 games just to show how much longer a game with little offense takes now than it did in the past.

The data can be filtered by innings, plate appearances and teams. In all cases, a game consists of both teams, which is necessary to show proper totals for individual teams. There are 2,430 games in every 162-game season, but since each game consists of two teams, they're double counted. Scrolling over the individual data points in the charts shows more information.

Play with the data as desired, but what you will find is that game length has increased dramatically no matter how it is controlled. For example, this chart shows the increase in game length as a function of total plate appearances for 2014 and 1988:

Across the board games were around fifteen minutes longer in 2014 than in 1988, even after controlling for innings and total plate appearances.

The reasons for this are well-known, including more mid-inning pitching changes, no real sense of urgency when a play is ended, players calling time not just at the plate but after every event on the base paths, replays that are taking a minimum of three minutes each, and there were 1,276 replay challenges in 2014, or about one in every other game. What these all have in common is that the average fan has zero interest in any of these things.

Rule 8.04 of the MLB Official Rules states:

8.04
When the bases are unoccupied, the pitcher shall deliver the ball to the batter within 12 seconds after he receives the ball. Each time the pitcher delays the game by violating this rule, the umpire shall call “Ball.” The 12-second timing starts when the pitcher is in possession of the ball and the batter is in the box, alert to the pitcher. The timing stops when the pitcher releases the ball.

The intent of this rule is to avoid unnecessary delays. The umpire shall insist that the catcher return the ball promptly to the pitcher, and that the pitcher take his position on the rubber promptly. Obvious delay by the pitcher should instantly be penalized by the umpire.

The more I think about it, excess time between pitches isn't the real issue. Of course, it is for some pitchers (and hitters), but an objective view shows this isn't really causing the problem. The increase in the number of strikeouts is reducing the number of balls in play, but that's a separate issue. It's the non-essential and inconsequential items that are adding time, and these can be removed without interrupting the flow of the game in any way. No one would notice, except they'd be more alert when watching.

Now, I have a confession to make -- when the NFL season started last fall, I was unable to watch baseball again until the playoffs. I just couldn't bring myself to watch games in which nothing was happening. Even though my team (the Cubs) brought up Javier Baez, Arismendy Alcantara and Jorge Soler and actually played very good baseball in August and September, the pace of the games simply wore me out. It's not really fair on my part, since the amount of activity in both sports is around the same, about fifteen minutes, but football just looks like more is going on.

I was talking with my 27-year-old nephew over Christmas, and he told me he's the only person his age he knows who watches baseball. The average age of baseball fans is almost 55, meaning I bring that average age down every time I tune in (for now). There's a graph in that previous link that should give Rob Manfred far more cause for alarm than infield shifts. It shows the average age of fans for nationally televised sports:

Sport	Median Age
NBA	39
NHL	45
NFL	48
NBA	55
Golf	58

I'm not an advocate of dumbing the game down or adding silliness to attract a younger audience, but there's no beauty in extraneous activity that adds nothing to the viewing experience. Keeping the games moving can get the game down to around two-and-a-half hours without sacrificing one second of ad time, so work to eliminate those extraneous behaviours that add additional time and no value. It can be done if there's a will to do it.

Data from Baseball-Reference

Scott Lindholm lives in Davenport, IA. Follow him on Twitter @ScottLindholm.