A Family Story
All of my children played either baseball or softball and I enjoyed watching their games over the seasons. However there was a sharp contrast in the game’s activity in my daughter’s slow-pitch softball game (when she was older) and my son’s fast-pitch baseball game (when he was young). In the softball game, there were many pitches put into play and there was a lot of fielding, runners on base, and runs scored. In contrast, my son’s baseball game consisted of young pitchers throwing to inexperienced young batters, the game consisted mainly of strikeouts and walks and it was rare to see a play in the field. It was more fun watching my daughter’s game.
Moving to professional baseball, there are similar general concerns about the pace of play, the increasing rate in strikeouts, and the small amount of time where balls are actually put into play. Here I explore this issue by exploring how the patterns of strikes, swings, and in-plays during a plate appearance have changed over the last 30 years of baseball.
I’m using Retrosheet play-by-play data for the seasons 1988 through 2016. I started looking at data for the 1988 season since this appears to the first season where Retrosheet kept track of different types of strikes such as called strikes, swinging strikes, etc. I suspect that the accuracy of this data may be questionable in the first seasons of recording this data but okay for more recent seasons.
Strikeouts and Walks
As we know the MLB strikeout rate has been dramatically increasing. The following graph shows the change in the K rate (K divided by PA’s) from 1988 through 2016. Note that we see an increase until some leveling off from 1998-2007, and the rate has been steadily increasing in the last 10 seasons.
In passing, note that the walk rate (BB divided by PA’s) has shown a different pattern over seasons. We see an increase from 1988 to about 2000, and it has been gradually decreasing in recent seasons.
Different Types of Strikes
Since one gets a K by getting three strikes on the batter, one would expect to see a historical trend in the proportion of strikes. There are four types of strikes during a PA, a called strike (CS), a swinging strike (SS), a foul strike (FS), and a ball put into play (IP). (By the way, I think it is convention to call a pitch put in-play a strike although the pitch’s location may not fall in the strike zone.)
Here we graph the fraction of strikes (CS + SS + FS + IP) / Number_of_Pitches and we see low values in the period 1988-2000, steady values between 2001 and 2010, and increasing values in the past six seasons.
How are the different types of strikes changing over time? Here I graph the fraction of called strikes CS / (CS + SS + FS + IP) over time. Interestingly, the fraction of called strikes increased and hit a peak about 2010 and has been dropping in recent seasons.
We see a different pattern in the fraction of swinging strikes SS / (CS + SS + FS + IP) over time. This fraction was pretty constant from 1988 through 2002, dropped to about 14% in the seasons 2003 through 2010, and has been steadily increasing in the last eight seasons.
Swings and Balls Put In-Play
Are batters swinging more? Let’s look at the fraction of pitches where the hitter swings. This is an interesting graph — I see a decrease from 1988 and a general leveling off from 1994 through 2012. There are four unusual seasons — 2001 and 2002 were unusually high, and 2009 and 2010 were unusually low. There has been a steady increase in the fraction of swings in the past four seasons.
Maybe it would be more meaningful to focus on the fraction of swinging strikes among all hitter swings. This fraction is defined by SS / (SS + FS + IP) where IP is the number of pitches put in-play. The pattern here is different — an increase from 1988 through 2002, a drop in the next six seasons (2003-2008) and a rapid increase in the last eight seasons.
Another way to look at this pattern is to look at the fraction of balls put in-play among all swings IP / (SS + FS + IP) — there is an obvious decreasing pattern. In 1988, about 45% of the swings were put into play and that percentage has dropped to about 39% in 2016.
What have we learned?
- Behind the historical rise in strikeout rates is a rise in the rate of strikes in a plate appearance.
- Although the rates of both called strikes and swinging strikes have generally been increasing over time, the rate of called strikes in the last two seasons has dropped.
- Batters are more likely to swing in a plate appearance, and the rate of missing a pitch in a swing has also increased.
- Since batters are more likely to swing and miss, the rate of balls in play on swings is dramatically decreasing.
Actually I got interested in this topic since I wanted to explore the home advantage effect in baseball. I believe much of the home advantage effect is to due to how umpires call balls and strikes and I thought Retrosheet data would be a good source of data to try to explore this home field effect. I’ll let you know how this exploration works out in a future post.
This is completely and utterly fascinating. I have been working on a project trying to explain this phenomena for the past four years. About a week ago I reached certain conclusions that your data echoes to a very high degree.
Additionally, I am a huge fan of your book “Curve Ball”, having read it a number of years ago and always wanted to thank you for your contributions to the growth of analytics in the best game ever.