Plate Discipline, Again

I looked at the concept of plate discipline a few weeks back, and I hypothesized that P/PA is not the best indicator of plate discipline (although it is the best existing measure) because of the fact that strikeouts add to the P/PA, and, theoretically, the most disciplined hitters would not strike out too often. I had the problem of weighting each part of this little stat that I was thinking about, but last night, at a very late hour, I had some ideas.

First and foremost, I decided to make it a sum of ranks, rather than an actual score. This was inspired by rotisserie-style fantasy baseball. That has its obvious flaws; Barry Bonds is not incredibly rewarded for his walks, but, in reality, Bonds's amazing walk totals are less a product of plate discipline than opponents' fear; he was intentionally walked a ton last year.

So this system essentially sums a player's ML ranks in lowest K/PA, highest P/PA, and highest (BB-IBB)/PA. There were 156 qualifiers for the comparison. The best score possible was a 3. No one got a 3 (A 3 would be #1 in all three of those categories). The worst score possible was a 468, I think, and no one got that, either. Here are some ranks, though.

1. Todd Helton - 59
2. Barry Bonds - 62
3. Johnny Damon - 70
4. Luis Castillo - 73
5. Scott Hatteberg - 75
6. Jason Kendall - 76
7. Rafael Palmeiro - 96
8. Ray Durham - 96
9. Bobby Abreu - 100
10. D'Angelo Jiminez - 110

It's an interesting spread of hitters, but most of them have a reputation for being very patient and disciplined hitters (especially a guy like Scott Hatteberg, whose OPS last year was one of the lowest among AL first basemen. You can see some additional value in this).

The worsts?

1. Corey Patterson - 420
2. Alex Gonzalez - 408
3. Pedro Feliz - 392
4. Geoff Jenkins - 374
5. Alfonso Soriano - 367
6. Aaron Rowand - 358
7t. Angel Berroa - 355
7t. Rocco Baldelli - 355
9. Torii Hunter - 347
10. Marquis Grissom - 344

Honorable mention goes to AJ Pierzynski, who scored dead last in two categories, but second in a third (he's tough to fan, but he doesn't walk or see pitches at all).

Plate discipline does not correlate with production, I don't think. While the trend line is negative in a graph comparing RC and plate discipline (as discipline decreases, so does production), the R2 value is only .14, which is not particularly significant, I don't think (I've never taken a stat class, so I'm trying to remember this from pre-calculus, calculus and high school math).

So now I have a plate discipline score to work with, which is what I was looking for. It's crude but it's another number to have around.

This is not the end, however.

I think that plate discipline is a skill that is not rated / scored correctly, even with a pretty simple little tool like this one. And I think I know how to fix that......there's a wealth of data out there that we don't keep, and it's important information. I want to know how a hitter responds to pitch location. I think that there are numbers to be made of a hitter's pitch selection. Does he tend to chase balls low and away? Does he not chase pitches at all? Does he foul off a lot of pitches? Does he swing at pitches out of the zone? What about if we take out the first pitch, which many batters take intuitively? A few sample numbers:

"Aggression frequency" - percentage of pitches considered balls that he swings at.
"Hack frequency" - percentage of balls that are substantially off the plate that he swings at.
"Contact frequency" - percentage of swings that make contact.
"Pitches taken frequency" - percentage of pitches taken.
"Called strikes frequency" - percentage of pitches taken for strikes.
"Called second and third strikes frequency" - percentage of pitches after the first strike taken for strikes.
"Protection frequency" - percentage of foul balls on two-strike swings.

I want numbers like that to evaluate plate discipline. Each number individually could be skewed (for instance, a player's high aggression frequency might be caused by the protection frequency), but, as a whole, they could tell us a lot. Players who are appearing at the top of the contact frequency list and the bottom of the hack frequency list would be extremely disciplined; they rarely miss when they swing and they don't swing at the bad ones too often. I don't think that these numbers are out there, but if they are, I'd love to see them.

I'm not exactly sure which numbers I would use if I were designing an evaluation tool....but I'm going to test it out (in the next game that I get to watch from start to finish) with my eye calling the balls and strikes when the batter swings, and letting the ump do it when the batter doesn't swing.

Anyway, the question is: how could you judge the balls and strikes? It occurred to me last night that you'd probably need some sort of computer system that was consistent on what it called balls and strikes to make your judgements really statistical, as opposed to the naked eye, which would be somewhat subjective. There is such a system though: Questec does just that. I don't know if they keep player records, but if the system could be used to keep player stats in addition to umpire stats, that would be outstanding.

So I'm going to give this idea a test run, probably in a weekend Met game, perhaps this Saturday night (for some reason, we get the WB11 on Gettysburg College's cable / satellite. Works for me). I'll sit with a pencil and paper with a scorecard of some sort to log all the pitches and locations, and then I'll crunch the numbers and see what it looks like.