A few weeks ago, umpire Joe West caused a stir when he publicly called out the New York Yankees and Boston Red Sox for doing what they do best: playing games that take forever to finish. He’s certainly not the first person (nor will he be the last) to wonder aloud why it is that baseball games last so long (particularly between those two teams). Is there a way to shorten the great American game? If there was, should we try?
In 2009, the average MLB game lasted 175.38 minutes, or just short of the three-hour mark. Contrary to popular belief, games played in National League stadia (that is, without the DH… unless it’s the All-Star Game) were slightly longer (by a couple of seconds) than games in American League stadia. Throwing out extra inning (and rain-shortened) games, the mark drops to 171.79 minutes, this time with AL games lasting a minute longer (172.31 to 171.32).
The quickest nine-inning game of 2009 was an interleague tilt between the White Sox (2) and Pirates (0) on May 22. It lasted 111 minutes (Gavin Floyd vs. Zach Duke, go figure). On April 25, two teams needed 4:21 to play an 8 ½ inning game (that ended 16-11). Anyone want to guess who the two teams were? As it happens, four of the 15 longest nine-inning games of 2009 were between the Yankees and Red Sox, and five of the other games in that group involved one of the two teams. In fairness, the Blue Jays, Indians, Mets, and Rays also made multiple appearances in that list as well. Perhaps some teams are simply built to be slowpokes?
During these discussions about shortening the game, the same set of culprits is identified over and over. Do managers really need to bring a reliever into the game mid-inning? Why don’t the umpires call the strike zone the way that it’s supposed to be called? What’s with all the throws to first? A few folks have floated proposals intended to shorten the game, but I have to wonder whether some of them are misguided. Leaving aside the counter-argument from addiction (a four-hour game is great because it’s baseball for four hours!), nerdiness (all the stuff that takes so long is the mental side of the game and that’s fascinating!) or sentimentality (it’s not often I get to spend much time with my dad, and I’d rather have four hours than two-and-a-half), there are a few logical counter-arguments to be made.
I’d argue that the much-maligned mid-inning pitching change might actually serve to shorten games. Consider that a reliever is generally brought into the middle of an inning for one of two reasons: to replace an ineffective pitcher or to gain a handedness match-up. The way that you make the clock move in baseball is to get batters to make outs like nervous seventh-graders at their first co-ed party. An ineffective pitcher isn’t recording outs, whether through fatigue or through just plain being awful. Managers try to play left-right match-ups because the platoon advantage is worth about 30 points of on-base percentage, also known as the “did you make an out or not?” stat.
What about throws to first? It’s true that it’s rare that a pitcher actually picks a runner off, but while a throw to first doesn’t make a runner less likely to try to steal, it does make him less likely to be successful when he does try. That means a greater chance of an out, and the greater chance that you’ll get home quicker.
But let’s see what the data say. I took all games from 2009, and ran a regression to predict the length of the game in minutes. I entered the following variables into the regression to see what shook out.
- The number of times a reliever was brought into the middle of an inning. A reliever brought into the game between innings can slip in and do his warming up during the usual time set aside for that, and the PA announcer can wait until after the fans are done singing Sweet Caroline to announce him.
- The number of throws to first made
- The number of pitches thrown
- The number of plate appearances for both teams
- The number of two-strike foul balls
- The number of stolen base attempts by both teams. After all, this is more than just an average pitch. There’s probably a throw down to second afterward, and that all takes time.
- The number of walks (intentional and unintentional considered separately), strikeouts, home runs, and balls in play hit by both teams
- Whether the DH was used (1 = yes, 0 = no)
- The percentage of pitches not swung at which were called strikes. This is a (very) rough estimate of the size of the strike zone, but obviously is conflated with the wildness of the pitchers involved in the game
- The number of between-inning breaks. Most of the time, that’s 16 or 17 for a nine-inning game, but of course in extra innings, there are more.
I ran the regression as a stepwise regression. For those not familiar, a stepwise regression looks first for which variable is the best predictor of the dependent variable (in this case, length of the game), and then runs the regression again looking for the next-best predictor, and so on.
The variable most associated with the length of the game was the number of pitches thrown. No surprise there. In fact, it picked up the first 82.3 percent of the variance. The next variable that entered was the number of mid-inning pitching changes, followed by the number of throws to first. Combined, those two picked up another 4.8 percent of the variance. Mid-inning pitching changes and throws to first really are a big driving force behind games being longer!
Several other variables entered the regression, although combined they only added another 2.1 percent to the R-squared. The final equation (for time of game, in minutes), for the morbidly curious was:
.430 * pitches + 3.084 * mid-inning relief + .673 * throws to first + 2.264 * IBB + .604 * plate appearances + .878 * SB attempts + .742 * inning breaks – .628 * HR – 1.362 * presence of DH – .154 * balls in play – 11.729.
The average mid-inning change of pitchers adds about three minutes to the game, while each throw to first adds about 40 seconds. In fairness, the average MLB game featured 2.06 mid-inning pitching changes and 7.28 throws to first. Even if they all disappeared, we’re talking about 10-12 minutes off the average game.
Other things of note included how much time an intentional walk and a stolen base attempt added to the game. Stolen bases, in addition to taking extra time beyond the average pitch, also indicate that the pitcher hasn’t been doing such a good job keeping runners off base. Intentional walks are apparently time intensive, probably because they occur during situations in which there’s a lot of “button-pushing” going on and everyone seems to be out of the dugout gesturing to each other. Then there are the events that apparently take negative amounts of time, including home runs and balls in play. The careful reader will notice that walks and strikeouts did not enter the equation, so what happens here is that each additional ball in play or home run probably comes at the expense of an additional walk or strikeout, both of which take more pitches to complete, on average. So, home runs and balls in play are a sign of a game where players are getting things over and done with.
There’s one variable that I can’t account for that probably has some influence. Some pitchers like to work quickly, while others take their sweet time. Rafael Betancourt, during his time with the Indians, earned the nickname Senor Slow Mo for his… reluctance… to throw the ball toward home plate. There may be some teams that either coach their pitchers to be slow (or fast) or simply just happen to employ a bunch of pitchers who work slowly (or quickly). However, there’s another lesson to be gained here. The main driver of the length of a game is how many pitches are thrown. More pitches means longer games, and teams who have batters who stretch out the count will play longer games.
Looking at the top 50 players in pitches per plate appearance for 2009 (minimum of 250 PA), we find Kevin Youkilis, Nick Swisher, David Ortiz, J.D. Drew, Brett Gardner, and Johnny Damon, all of whom played for either the Red Sox or Yankees last year. Also in the top 50 were Nick Johnson, Jeremy Hermida, and Marco Scutaro, now members of the Yankees and/or Red Sox. It looks like the Yankees and Red Sox have put their teams together and coached their players, intentionally or unintentionally, to try to win games by a war of attrition. (I should note that I’m not the first person to notice this.) It’s not really a bad strategy. If you can tire the starter out before the sixth inning, you often get to see the soft underbelly of the other team’s bullpen. Even within an inning, even at the same pitch count, there’s probably a difference between throwing my 70th pitch, and it’s the 20th of the inning versus it being thrown after I’ve had a few minutes to sit and rest in the dugout between innings (although I’ve not seen any research addressing the topic.) The problem with this war of attrition strategy is that it produces games that take three-and-a-half hours.
So if baseball really wants to shorten up games, it could cut 15 minutes out per game if it cut out two pitches of each half inning. How to encourage that? Well, consider the circumstances under which the war of attrition strategy has flourished. Pitchers go out to the mound with a sign around their neck that says “I’m leaving after 100 pitches.” Teams also appear to unflinchingly stick to an inning-based usage schedule for relievers. If it’s the sixth inning, then Smith is going to be the guy to get the call, never mind that Smith is awful and the game is tied. He’s our sixth-inning guy. Given that, it makes sense to try to stretch the starter out a little bit in the hopes that you can exploit this quirk of what is loosely called “strategy.”
Baseball has inflicted this on itself. The merits of the 100 pitches-and-out rule can be debated, but it’s silly to expect that as this has become more emphasized that the game wouldn’t evolve along with it. If baseball wants shorter games, it can toy with little rule changes, but what it’s really fighting is the natural evolution of the game, and that’s hard to do.
—
And just as soon as it began, it was over. This article is my last one for BP. It’s sad because I will miss all the folks who read and commented on my stuff on the page and behind the scenes. So if you are reading this, thank you. I want to thank Kevin Goldstein, Christina Kahrl, John Perrotto, and all my co-workers at BP who have been a treat to work with. Special thanks go out to Eric Seidman who when I mentioned back in November that I was interested in getting back into writing about baseball asked “Wanna write for BP?”
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
***
One of my issues with regression at this granularity level is when I see something like this: ".742 * inning breaks "
Well, we know that inning breaks are 2 to 3 minutes each, depending which TV network is involved. So, what the regression is saying is that there's some 1.5 to 2.0 minutes that it's removing from what we know, and distributing it to other variables, even though, in this particular case, it should be completely independent. That is, the between inning break has no relationship whatsoever to any other event. But the regression is finding some relationship.
***
Cutting one minute in the non-action between inning will save some 17 minutes of game time. The players loaf around too much by their own admission. But, as one of the players recently admitted on his blog "we got to gets paid".
So, this is really the issue: how can you cut down on game time while not touching the non-game time. Which is a very weird thing to try to optimize from a fan experience.
Indeed, what's to stop MLB from increasing between inning game time, even if we reduce the actual game-time, so that we are always going up the same hill? Sisyphus anyone?
"Mid-inning pitching changes and throws to first really are a big driving force behind games being longer!"
Maybe - but both of those are symptoms of runners on-base. No where on this list is "runs scored" or "closeness of game", both of which would also seem to correlate well with runners reaching base. Yes, there was plate appearances, but I would think the closeness of the game would matter, too. Similarly, if you allow a bunch of runners to score after reaching base, perhaps the game goes faster than if you strand all those runners?
Thanks for the memories, Russell.
One quibble. It has been said before, but bears repeating that although West and others have complained about the length of games, what they really mean is the pace of games. I don't think people are upset with 4 hour games that feature lots of action; they don't like 3 hour games that drag because of trips to the mound, batters leaving the batter's box, pitchers meditating on the mound and so on.
So the solution, if there is any consensus that the pace needs to be picked up, is not concerned with getting quicker outs but moving the action along.
I do want to note that I am not in the least bothered by the slow pace of games or their length but respect the opinions of those who are.
This article brought to mind an issue I have been ruminating on for much of the early season - the pace of baseball is much more leisurely than that of hockey - my winter game. It's a nice change, and a much more pleasant way of passing the summer months. I am much more annoyed at the time spent on non-baseball events than I am by the pace of the game on the field. I would love to see baseball experiment with radical time saving measures during sprint training - no between inning breaks, for instance, or limiting the number of non-injury-related mid-inning pitching changes, or enforcing a much quicker time limit per pitch.
Good luck in the future, Russell - I hope your reasons for leaving are ones you are happy with. All the best.
I looked at the scoreboard to check the count after a pitch and missed the next one. Gibson would get the ball back from the catcher, take his sign and fire away.
Now we have pitchers walking around behind the mound and batters stepping out of the batter's box after every pitch to readjust their batting gloves.
It should be really simple to speed things up, especially when there are no baserunners, by enforcing two rules; the batter must stay in the batter's box and the pitcher has a certain amount of time to make a pitch.
As for shortening time between innings, it's not gonna happen as long as companies are paying for commercial time.
It seems like the most obvious rule change suggested is an old suggestion, the 'let em go' -- essentially pointing to first base and saying 'take your walk'. It would seem for the <1% change that 'something might happen' (wild pitch, swing at a ball too close), that the benefit of the faster game through (to me) a tedious and boring 2 to 4 minutes would be a net benefit.
One major missing variable- nationally televised game vs. not nationally televised. There are about 30% more commercials in nationally televised games.
All this being said- great article!!!
- The slow pitcher. Betancourt actually had the clock called on him a few times in Cleveland, resulting in the ball/whatever penalty when there's no men on base.
- The slow batter. The guys that step out of the box and/or fidget around for 30 seconds between pitches.
- The "Time outs". Batters calling time when the pitcher takes too long to start his motion, as well as pitchers stepping off to recycle the signs.
I don't want to regulate anything that really impacts gameplay. You need to be able to make pitching changes mid-inning, and you can't limit the # of pickoff throws (why wouldn't the runner take a huge lead once the limit is hit?).
If there are (random guess) 300 pitches per game, and you can shave 2 seconds off the time between pitches, you just cut 10 minutes out of the game. It seems like it wouldn't be too hard to do that by policing the things I mentioned above.
Russell, going to miss your odd angles.
I'm sure they announce the call to the batter verbally first, but the delayed point to signal a strike, drives me insane as a viewer. It seems like I've seen it more this year than any other year.
Okay, enough of me whining...