August 7, 2007
Non-Contact Part II: More on Strikeouts
If you remember, last time out we looked at that sabermetric darling, the strikeout, noting the steady increase in strikeout rate that we've seen over the past four decades. In part two, we'll look at a couple of factors that affect pitcher's strikeout rates.
One factor sometimes ignored when considering a pitcher's (or hitter's) strikeout rate is the effect that the various major league ballparks can have on the strikeout rates of the hurlers who pitch in them. We're familiar with the concept of a ballpark effect when we talk about offense: the potential home run-boosting effect caused by close outfield fences, or of the possible hit-suppressing effect of a large area of foul ground on the field of play. These have an effect on batted balls that we can understand intuitively. Less intuitive is that the ballpark can also increase or reduce the rate of non-contact results like walks and strikeouts.
Why does this happen? Much of the ballpark effect on strikeouts is attributed to visibility factors. First, no two ballparks have an identical "batter's eye," the background against which the batter sees the pitch after it leaves the pitcher's hand. There are other factors, however, like the late afternoon shadows at ballparks like Wrigley Field, which are also thought to have an effect on how well the batter sees the pitch, and accordingly how likely the hitter is to get rung up on strikes. High altitude makes it more difficult to throw breaking pitches, a factor which is thought to be responsible for the extreme strikeout-suppressing qualities of Colorado's Coors Field. It is possible that other elements, such as how an individual park's groundskeepers maintain the pitching mound, could add or detract from a pitcher's effectiveness, including their strikeout rate.
Here's a list of the ballpark strikeout factor for each team's home field. The strikeout factor is arranged so that 1000 is a perfect strikeout-neutral ballpark; higher numbers mean the park increases strikeouts, and lower numbers mean that the park suppresses strikeouts.
Team Park Factor Team Park Factor SEA Safeco Field 1047 CIN Great American Ballpark 1002 FLO Joe Robbie Stadium 1040 LAA Anaheim Stadium 998 MIL Miller Park 1035 NYA Yankee Stadium II 997 SDN Petco Park 1034 NYN Shea Stadium 993 MIN Metrodome 1026 PHI Citizens Bank Park 988 CHN Wrigley Field 1017 TEX Ballpark at Arlington 987 TBA Tropicana Field 1015 SFN PacBell Park 987 CLE Jacobs Field 1014 SLN New Busch Stadium 985 LAN Dodger Stadium 1012 ARI Bank One Ballpark 978 TOR Skydome 1011 KCA Kauffmann Stadium (04-06) 978 ATL Turner Field 1008 OAK Oakland Coliseum 975 WAS RFK Stadium (Nationals) 1008 PIT PNC Park 974 BOS Fenway Park 1004 DET Comerica Park (03-06) 973 CHA U.S. Cellular (01-06) 1003 BAL Camden Yards 972 HOU Minute Maid Park 1003 COL Coors Field 937
Reader J.D. chimed in with this comment after my first strikeout article:
I liked your article on strikeout rates, but I think that in addition to adjusting for eras there is another bias that favors modern pitchers. Today's pitchers get pulled in the sixth inning at about 110 pitches, no matter how they're feeling. Naturally they will have higher strikeout rates than earlier pitchers who had labor through nine innings, seeing their strikeout rates diminish due to fatigue in later innings. I don't know how you can adjust for this, but I think that it does benefit modern pitchers beyond just the overall strikeout rates of the modern era. I think Sandy Koufax averaging 10 strikeouts while also averaging close to nine innings is just good, if not better, than Kerry Wood averaging eight strikeouts in seven innings. Like you said right off the top of the article, strikeouts per nine innings is antiquated since nobody goes nine innings anymore.
Now, underlying J.D.'s question is an assumption that strikeout rate goes down as pitchers throw more innings/pitches. This was mentioned in a number of letters I received, with reader D.K. asking point-blank, " Do starting pitchers have more strikeouts when their pitch counts are lower or higher?"
This seemed like a very good question, so I did what any reasonable person would do: I asked someone much smarter than me-in this case, the indefatigable Bil Burke-to show me some data.
These numbers are based on pitcher's individual pitch counts, tied to the number of pitches thrown at the end of each plate appearance. What the numbers show is a clear downward trend in strikeout rate after the first 50 pitches, presumably as pitchers tire. After 100 pitches, the strikeout rates go up again, as the pitchers to surpass the 100-pitch barrier are generally pitching well, but also get pulled promptly if they run into any trouble.
There's also an issue of selection bias, where better pitchers with good strikeout rates are allowed to pitch deeper into games. If you want to get a pretty good idea of the quality of the pitchers who are going over that 100-pitch mark, you can look at the Pitcher Abuse Point report for the last eight seasons, where pitcher's starts are categorized according to the number of pitches thrown per start. When you look at Categories 4 and 5, which represent pitch counts of 122 and higher, you can see that the pitchers who allowed to pitch that deep into the game are a fairly distinguished group-you see a lot of Randy Johnson and Curt Schilling on those lists-so it makes sense that the strikeout rates remain higher than average in the last three categories.
Getting back to J.D.'s original question- Sandy Koufax versus Kerry Wood, and the question of whether or not the historical rate adjustments done in the Davenport Translations capture the differences between the eras in which they pitched-we see that J.D. might have a point. Quicker hooks and attention to pitch counts for modern-day starting pitchers mean that they can pitch in an all-out fashion, knowing that they're not expected to complete the game, while pitchers in Koufax's era are generally considered to have paced themselves in order to go the distance.
But a closer look at the data above indicates that perhaps the increases in strikeout rates of the 21st century aren't being driven by the starting pitchers like Kerry Wood, but rather by relievers. The starter's strikeout rate from 2000-2007 is 15.88 percent; for relievers, it's 18.44 percent. These high-strikeout relievers are pitching a greater percentage of team innings than they did in Koufax's time. For example, in 1961 relievers pitched 28.5 percent of their team's innings, by 2001, the reliever's share of innings had risen to 33.4 percent. Since the Davenport Translations are based on league strikeout rates, it's possible that a modern starter would be underrated, because relievers are inflating the league strikeout rate.
We'll continue this discussion in part three of the Non-Contact series, where we'll talk some more about strikeouts, and add hitters to the mix.
Dayn Perry, "Can of Corn-Putting the Park Back in Park Factors": An introduction to the concept of component park factors.
James Click, "Crooked Numbers-Ballpark Effects on Pitcher Types": Looks at the effect of the league-wide ballpark factor on the performance-including strikeout rates-of fly ball and ground ball pitchers.
Nate Silver, "Lies, Damned Lies-What to Do With Papelbon": A study of the effects of converting starting pitchers to relief, with the finding that starters so converted will generally have significantly higher strikeout rates than they would as relievers.