Baseball Musings
Baseball Musings
April 20, 2006
Pitchers and Pitches Per PA

The Baseball Savant doesn't buy the idea that fewer innings pitched per start is mostly the result of more selective batters seeing more pitches per plate appearance.


Posted by David Pinto at 12:25 PM | Statistics | TrackBack (0)
Comments

I don't buy his argument.

My understanding is that the real problem is the pitches you throw when you are tired. Anecdotally, I have noticed that long at bat innings often take something out of a pitcher.

Combining the two, I can see 2 mechanisms where a small per PA increase leads to a shorter total pitch count per game.

A 0.2 increase in #P/PA leads to about a 1 pitch per inning increase. The last pitch in an inning is gong to be tougher than any other in that inning, so extending every inning by 1 pitch could be harmful.

I also wonder if there are more 8+ pitch plate appearances now. If the difference now is 4 long plate appearances a game that add 2-3 pitches apiece, I can easily imagine that having a significant impact.

Posted by: Craig A. Damon at April 20, 2006 04:38 PM

Craig,

Great comment. That is something I thought about too actually and when I finished posting my comment, that is the one thing I wish I knew more about. Is the question really about total pitches? That is sort of why I wrote the article, in that I think you can't chalk up fewer innings to pitches per plate appearance in total because you're essentially arguing total pitches. But what is the alternative? If you think an 11-pitch at-bat kills a pitcher or a 30 pitch inning kills a pitcher, when is it best to yank him out of the game? What if a pitcher goes through a game and throws an 80-pitch complete game shutout but throws 30 pitches in the 2nd inning in which he walks a couple of batters and goes full count on everyone? Is the manager wise to pull the pitcher after his 20th pitch in the 2nd inning? Can we really know?

Posted by: Baseball Savant at April 20, 2006 05:51 PM

Craig's suggestion is intriguing, and I believe it could be a bit of the explanation. I've wondered myself whether the lengthening of games via increased time between pitches, and increased time between innings has something to do with it. There might be more cumulative fatigue felt on the 100th pitch delivered 2 hours and 45 minutes into the game than on a 100th pitch delivered after only two hours in a quicker game.

Another way to look at the problem is not by looking at league averages (quality or preferred style of pitchers used may change over time) but instead to look at individual pitchers' progress.

For instance, an examination of Roger Clemens' career shows no clear trend in increasing pitch counts, measured by pitch per PA, or per IP. His TBF per start and pitches per start didn't drop until he joined the yankees in 1999. That could be coincidence and the impact of age, but it could also be the dynamics of the staff: great closer, dependable setup men, reliable fellow starters so that he wasn't under pressure to pitch another inning and give the bullpen a rest...

There's a common thought (I think) that pitching staffs went to 11-12 men because starters couldn't go as long. But perhaps cause and effect have been reversed - starters don't go as long because the bullpen is full of lefty/righty specialists who need work. There's a greater tendency to remove starters before they tire rather than when they tire.
Here's a chart for Clemens using ESPN data:

#PIT TBF #P/PA #P/IP #P/GS ip/GS tbf/gs
1987 Bos 4271 1157 3.69 15.2 119 7.9 32.1
1988 Bos 4186 1063 3.94 15.9 120 7.5 30.4
1989 Bos 4243 1044 4.06 16.7 121 7.2 29.8
1990 Bos 3530 920 3.84 15.5 114 7.4 29.7
1991 Bos 4031 1077 3.74 14.9 115 7.8 30.8
1992 Bos 3821 989 3.86 15.5 119 7.7 30.9
1993 Bos 3170 808 3.92 16.5 109 6.6 27.9
1994 Bos 2895 692 4.18 17 121 7.2 28.8
1995 Bos 2454 623 3.94 17.5 107 6.1 27.1
1996 Bos 4260 1032 4.13 17.6 125 7.2 30.4
1997 Tor 4106 1044 3.93 15.6 121 7.8 30.7
1998 Tor 3806 961 3.96 16.2 115 7.1 29.1
1999 NYY 3327 822 4.05 17.7 111 6.3 27.4
2000 NYY 3433 878 3.91 16.8 107 6.4 27.4
2001 NYY 3604 918 3.93 16.4 109 6.7 27.8
2002 NYY 2995 768 3.9 16.6 103 6.2 26.5
2003 NYY 3460 878 3.94 16.3 105 6.4 26.6
2004 Hou 3431 878 3.91 16 104 6.5 26.6
2005 Hou 3203 838 3.82 15.2 100 6.6 26.2

Posted by: joe arthur at April 20, 2006 06:35 PM

Joe,

I can buy that argument. That would seem odd though when you think about how much money the market has thrown at pitchers in recent years. If there is a reverse in cause and effect and pitchers are now being pulled before they are tired, then wouldn't this then shift the market to imbalance and we could be on the verge of a Moneyball effect in that the team that figures out that pitchers really can throw 275IP could be a decisive advantage? Take a look at BJ Ryan from Toronto. If the guys is making 11 million dollars and yet just pitching 60 innings then he's pitching about 6% of the innings the Blue Jays will face. That is a tremendous market inefficiency when your alternatives could be Vinnie Chulk or Justin Speier. The same goes for starters.

Intuitively something doesn't feel right about it. With the advancements of player development and health, why are we supposed to believe that hitters are so much better, but pitchers have become more fragile than our grandmother's china?

Posted by: Baseball Savant at April 20, 2006 09:19 PM

Savant,
it could be that pitchers are closer to the edge than ever before; they have to throw harder and throw more types of pitches to be competitive; maybe these stresses are balancing the health improvements. Certainly there's been discussion recently that closers are currently used in a suboptimal way; it could be that top starters are used suboptimally too.

Anyway, I should have looked at some data before going off on a tangent originally. To address the original theory: Yes, a rise of .21 pitches per plate appearance since 1988 would mean a pitcher would reach eg 100 pitches in 1.5 fewer batters faced. What's 1.5 batters faced? About 1/3 of an inning pitched. Yes, that seems fairly trivial in terms of explaing the steep decline in complete games, or the decline in innings by top starters.

However, it does match the actual decline in average innings pitched by all starters over the years in question. Using retrosheet data, in 1987-89, AL starters averaged 6.20 innings and 26.54 BFP/start. In 2002-4, AL starters averaged 5.93 IP/start and 25.66 BFP/start. [I use AL only because the DH means there's no side effects from pitcher's batting skills or need to be removed for pinch hitters before they're fatigued.] So if the question has to do with the average starter, Woolner's explanation seems to work pretty well.

Posted by: joe arthur at April 20, 2006 10:03 PM
Post a comment









Remember personal info?