from Amazon
Wednesday, January 23, 2008
Does PAP work?
Beats me. I did a mini-test. I started off with the PAP at BP’s site.
I selected the 90 pitchers with the most pitches thrown in 2005, reasoning that these pitchers would reasonably be expected to be pitching in 2006 based purely on talent. From those 90 pitchers, I selected the 30 pitchers with the most “stress” (defined by BP as PAP points divided by pitches thrown), and the 30 pitchers with the lowest stress, and the 30 middle pitchers.
I was interested to see how many pitches each group threw in 2006.
The 30 highly stressed pitchers in 2005 threw 3325 pitches in 2005 (on 32.4 starts), and followed that with 2961 pitches in 2006, or an 11% drop. The 30 least-stressed pitchers in 2005 threw 2984 pitches (on 32.0 starts), and followed that up with 2176 pitches, or a 27% drop! Ouch. Sounds like we want our pitchers stressed.
Maybe there’s some quality check that we can do here. Maybe the guys who are being stressed are also the better pitchers. Let’s redo, but this time only look at the top 60 pitchers in pitch counts. This time, the high-stressed pitchers averaged 3400 pitches on 32.7 starts, and followed that up with 3100 pitches in 2006. The low-stressed pitchers averaged 3150 pitches on 33.1 starts, and ended up with 2368 pitches in 2006. Once again, the more stressed, the better.
Let’s repeat by selecting pitchers in 2006, and seeing how they did in 2007. The 30 high-stressed pitchers of 2006 averaged 3256 pitches on 31.8 starts, and ended up with 2866 pitches in 2007. That’s a 12% drop. And the 30 low-stressed pitchers of 2006 averaged 2923 pitches on 31.0 starts, and ended up with 2255 pitches, or a drop of 23%.
Repeating with only the top 60 pitchers in pitch counts, instead of 90: top 20 stressed pitchers had 3271 pitches on 31.7 starts in 2006, followed by 3002 pitches in 2007. The 20 least-stressed pitchers had 2982 pitches on 31.3 starts and followed that up with 2267 pitches.
After I did all that, I looked at the pitchers and saw “Cory Lidle” and “Brad Radke”. To exclude pitchers like that, I put in a requirement of at least 1 pitch thrown in the subsequent season. This will of course remove pitchers who were genuinely hurt all season, but someone else can pick up the slack here.
Anyway, repeating all that work, here are the results:
2006 pitchers, n=90
high-stress NP=3256, GS=31.8, in 2007 NP=2866, 12% drop
low-stress NP=2920, GS=30.9, in 2007 NP=2393, 18% drop
2005 pitchers, n=90
high-stress NP=3325, GS=32.4, in 2006 NP=2961, 11% drop
low-stress NP=2969, GS=31.9, in 2006 NP=2280, 23% drop
2006 pitchers, n=60
high-stress NP=3430, GS=33.0, in 2007 NP=3062, 11% drop
low-stress NP=3189, GS=32.7, in 2007 NP=2701, 15% drop
2005 pitchers, n=60
high-stress NP=3400, GS=32.7, in 2007 NP=3100, 9% drop
low-stress NP=3144, GS=33.0, in 2007 NP=2429, 23% drop
The last two charts I prefer, as they are with n=60 (60 top pitchers in pitch counts). There should be very little quality seepage here. If you are throwing at least 3000 pitches in a season (the rough baseline), you’ve got to be considered a good pitcher.
Anyway, if you remember in my Buehrle/Santana discussion, I said top pitchers in their late 20s, THROUGHOUT HISTORY, will lose about 10% of their IP per year into their 30s. And this is without account for any “stress” activity. And the above charts, when focusing on “stressed” pitchers, also show a 10% drop.
Even if you want to argue that the low-stress pitchers include alot of subpar pitchers (remember, there’s only 60 starters in the group, meaning that you think that their team will allow them to be one of the top 2 starters on their team, and that the next year NO team would think that they were one of the top 150 starters), and want to disregard the 15% and 20% drops, what are you going to argue as the control? That the high-stressed pitchers should have only lost 5% of their pitch counts, but the “stress” caused them to lose 10%? But, as I’ve shown, we should expect the average top pitcher, at any point in history, to lose 10%.
It seems to me that this stress business is not what it’s all about. Does PAP work? I don’t know that it does. But, it’s not for me to prove that it doesn’t work. It’s up to someone else to prove that it does work. And so far, I don’t see it.
The top 30 pitchers had an ERA of 3.53; the 61-90 pitchers had an ERA of 4.63. The middle 30 were at 3.95. So the quality leekage may be an issue.
It seems that we should control for both pitcher quality and pitcher age, and determine how many innings we would “expect” from each pitcher the following season. Does Marcel do this? We may find that “stressed” pitchers underperform their Marcel IP more than the “non-stressed” (not that I think this would happen, actually).
Interestingly, the top 30 pitchers don’t just throw more pitches because they’re better. The top 30 also averaged 3.78 pitches per batter faced, the middle group 3.69, the bottom 30 3.62. This is probably because the top group K’s more batters—19.5% of total batters faced, where the middle group’s at 16.5% and the bottom at 13.5%. (The top group also walked a bit more, 7.2% to 6.8% to 6.7%.)
The pitches per inning is somewhat interesting. The top group averaged 15.76, the middle 15.56, the bottom 15.65. If we assume that managers start to mull over pulling their SP at the 100-pitch mark, we would expect the top group to hit that mark after 6.35 innings, the middle after 6.43, and the bottom after 6.39. In actuality, the top group averaged 6.60 innings per game, the middle 6.26, the bottom 5.82. (A few of the pitchers had some relief appearances, I don’t know if that’s throwing things off.) The bottom 30 were on a really short leash with their managers, apparently.