Wednesday, March 19, 2008

Batter P/PA Since 2000

A lot of the lineup talk in Milwaukee these days focuses on Jason Kendall batting ninth, and perhaps rightly so. After all, it's unfamiliar for most baseball fans to see a pitcher bat above ninth; it seems to make little sense to have your worst hitters at bat more during the season. Plenty of people have written about why it makes sense to put Kendall in the ninth spot and there's no need for another such post here. Instead, I want to focus on another part of the Brewers lineup and part of the reasoning therein.

New center fielder Mike Cameron is slated to bat second in the regular lineup once he returns from a 25-game suspension in late April. The team points to his ability to work counts and wear down opposing pitchers by seeing over four pitches per plate appearance (P/PA). In a related vein, fans last year were often frustrated by erstwhile catcher Johnny Estrada's tendency to swing at the first pitch, leading to a very low 2.99 P/PA mark last season. has data including the number of pitches faced by every batter since 2000. Well, actually, it goes before that but the data seems to get less reliable (some PA missing, etc.) the further back you get. I used 2000 as a cutoff point because that's the first year used by the Play Index when cataloging pitch data for pitchers.

Four hundred and seventy-four batters have had 1000 or more plate appearances since 2000. The average plate appearance by those 474 batters lasted 3.76 pitches with a standard deviation of 0.231. The highest number of P/PA seen was the 4.50 by Jayson Werth. The lowest was the 3.04 seen by former first baseman Randall Simon. Perhaps most interesting, however, is that it really doesn't matter how many pitches you see in an average plate appearance: it has little bearing on the common rate statistics: batting average, on base percentage, and slugging percentage.

For example, below is a chart showing batting average against P/PA for all batters with 1000+ PA since 2000. You will note the data is essentially random: the r-squared value is 0.0239.

(Click to enlarge in a new window)

The chart for slugging percentage can be found here; that data has an r-squared value of .0436. On base percentage has the "strongest" correlation (see the chart), but even the r-squared of that data is only .2041. Of course, it makes sense that seeing more pitches lets you increase your on base percentage: you need to see at least four pitches to draw a base on balls, usually a major part of a high on base percentage.

So perhaps the number of pitches per plate appearance really doesn't impact a batter's individual numbers. It's certainly possible it can wear down a starting pitcher faster, but I tend to question the advantage of, say, only one more pitch every three or four plate appearances (Cameron's 4.07 P/PA is 0.31 pitches over the average).

Below are lists of the top and bottom players on the active lists.

Most P/PA Among Active Players, Minimum 1000 PA, 2000-2007
  1. Jayson Werth, 4.50
  2. Kevin Youkilis, 4.40
  3. Bobby Abreu, 4.32
  4. Brad Wilkerson, 4.27
  5. Adam Dunn, 4.24
  6. Jason Giambi, 4.20
  7. Nick Johnson, 4.20
  8. Mark Bellhorn, 4.19
  9. Pat Burrell, 4.19
  10. Frank Thomas, 4.19
  11. Orlando Palmeiro, 4.17
  12. Nick Swisher, 4.17
  13. Jim Thome, 4.16
  14. Casey Blake, 4.14
  15. Jim Edmonds, 4.14
  16. Jose Bautista, 4.14
  17. Dan Johnson, 4.13
  18. Troy Glaus, 4.12
  19. Jamey Carroll, 4.12
  20. Gregg Zaun, 4.10
Least P/PA Among Active Players, Minimum 1000 PA, 2000-2007
  1. Nomar Garciaparra, 3.16
  2. Yuniesky Betancourt, 3.22
  3. Johnny Estrada, 3.23
  4. Robinson Cano, 3.24
  5. Vladimir Guerrero, 3.26
  6. Neifi Perez, 3.29
  7. Jay Payton, 3.30
  8. A.J. Pierzynski, 3.30
  9. Sandy Alomar Jr., 3.31
  10. Toby Hall, 3.33
  11. Garret Anderson, 3.33
  12. Corey Patterson, 3.34
  13. Juan Castro, 3.34
  14. Jason Tyner, 3.34
  15. Cristian Guzman, 3.35
  16. Bengie Molina, 3.35
  17. Alex Cintron, 3.35
  18. Pedro Feliz, 3.35
  19. Placido Polanco, 3.36
  20. Kenji Johjima, 3.38


Anonymous said...

it does get the starting pitcher out of the game a whole lot faster though.

If every batter saw 5 pitches per plate appearance and the pitcher allowed one base-runner an inning (WHIP 1.00), that would be 20 pitches per inning.

the pitcher would hit 100 at the fifth inning and in come bullpens which usually stink

TheJay said...

True enough, though the average start only lasts about 5 2/3 or 6 innings anyway, so you're not gaining too much, especially if you sacrifice on base percentage in favor of seeing more pitches before making outs.

Marty said...

Those are interesting lists. The top P/PA tend to be good hitters, and the bottom are both great hitters (vlad) and awful hitters (neifi). I expect that would hold up under further exploration - high p/pa guys are probably good hitters, but low p/pa tells us little.

Also, for anonymous, when you're talking about a stat where the top guy sees 4.5 p/pa, hypothesizing a whole team of guys seeing 5 p/pa is not very useful.

Anonymous said...

But you are gaining a lot because the average starter has a WHIP much higher than 1.00

Marty said...

I would say that using heavily skewed pitching numbers to make up for heavily skewed batting numbers is going lead to conclusions with a tenuous relationship to reality :)

Charlie said...

Where can I get the raw data for this?

I looked at BR and couldnt track it down.

Theron Schultz said...

BR has it for individual hitters under the Special Batting information under Pitch Data Summary. I used the Play Index to get a list of every batter with 500+ PA since 2000 and had to put all the numbers together by hand. I have an Excel file of the data - just send me an email if you're interested.