Today, I’ll continue with the batted balls versus pitchers theme which I started earlier this week. An at bat can result in any of the following events:
- Strikeout
- Base on balls
- Hits batsman
- Ground ball
- Line drive
- Outfield fly
- Infield fly
Three of those events are generally favorable events for pitchers:
- Strikeout
- Ground ball
- Infield fly
I will call these run preventing events (RPE). Of course, a ground ball is not as easy an out as a strikeout or an infield fly and can have a negative result for a pitcher. However, inducing a lot of ground balls will help to prevent runs over the course of a season. On the other hand, it is good for pitchers to avoid, for the most part, the following events:
- Base on balls
- Hits batsman
- Line drive
- Outfield fly
So, I thought it would be interesting to calculate the run preventing event percentage (RPE%) for American League pitchers in 2006 where RPE% = (SO + GB + IFF)/BFP. Striking out batters and inducing grounders have been shown to be repeatable skills. Getting batters to hit infield flies is not stable from year to year (correlation = .10 between 2005 and 2006). However, infield flies are relatively rare compared to other batted ball types and including them does not change the RPE% substantially in most cases. Plus, I suspect (without statistical evidence) that this is a real ability for some power pitchers.
It turns out that RPE% is fairly stable with a .66 correlation between 2005 and 2006. It can also be considered a fielding independent stat because, although the end result is not independent of fielders, getting a grounder or infield fly to happen in the first place has nothing to do with fielders. It is as stable or more stable than FIP ERA but it is not weighted and thus does not explain as much about runs allowed. RPE% is just an exploratory stat for now.
There were 48 American League starters with 125 or more innings pitched in 2006. Table 1 lists the RPE% rankings for Tigers starters. Table 2 lists the top 20 pitchers in the league. We can see that Jeremy Bonderman (RPE%=.585) ranks very well (6th in the
Table 1: Run Preventing Events for Tigers Starters in 2006
Name | BFP | SO | GB | IF | RPE | RPE% | |
6 | Bonderman | 903 | 202 | 306 | 20 | 528 | .585 |
14 | | 849 | 99 | 340 | 18 | 457 | .538 |
16 | Robertson | 881 | 137 | 312 | 23 | 472 | .536 |
27 | Verlander | 776 | 124 | 244 | 25 | 393 | .506 |
--- | Ledezma | 264 | 39 | 67 | 15 | 121 | .458 |
--- | Maroth | 234 | 24 | 81 | 4 | 109 | .466 |
--- | Miner | 398 | 59 | 144 | 8 | 211 | .530 |
Table 2: Top 20
Rank | Name | Team | BFP | SO | GB | IF | RPE | RPE% |
1 | Halladay | TOR | 876 | 132 | 404 | 22 | 558 | .637 |
2 | Hernandez | SEA | 816 | 176 | 331 | 10 | 517 | .634 |
3 | Wang | NYA | 900 | 76 | 484 | 7 | 567 | .630 |
4 | Westbrook | CLE | 904 | 109 | 447 | 7 | 563 | .623 |
5 | Burnett | TOR | 577 | 118 | 208 | 14 | 340 | .589 |
6 | Bonderman | DET | 903 | 202 | 306 | 20 | 528 | .585 |
7 | Santana | MIN | 923 | 245 | 255 | 32 | 532 | .576 |
8 | Sabathia | CLE | 802 | 172 | 261 | 26 | 459 | .572 |
9 | Mussina | NYA | 804 | 172 | 251 | 36 | 459 | .571 |
10 | Bedard | BAL | 844 | 171 | 292 | 14 | 477 | .565 |
11 | Kazmir | TB | 610 | 163 | 165 | 14 | 342 | .561 |
12 | Haren | OAK | 930 | 176 | 316 | 29 | 521 | .560 |
13 | Schilling | BOS | 834 | 183 | 247 | 28 | 458 | .549 |
14 | | DET | 849 | 99 | 340 | 18 | 457 | .538 |
15 | Escobar | LAA | 789 | 147 | 263 | 13 | 423 | .536 |
16 | Robertson | DET | 881 | 137 | 312 | 23 | 472 | .536 |
17 | Contreras | CHA | 833 | 134 | 283 | 29 | 446 | .535 |
18 | Johnson | NYA | 860 | 172 | 258 | 30 | 460 | .535 |
19 | Millwood | | 907 | 157 | 309 | 18 | 484 | .534 |
20 | Beckett | BOS | 868 | 158 | 282 | 23 | 463 | .533 |
This is awesome stuff Lee. I like the simplicity of the stat in that it's simply looking at a rate for good stuff versus bad stuff.
ReplyDeleteI'm assuming that it correlates with runs allowed pretty well?
Bill, it has a moderate correlation with runs scored: -.50 for 83 AL pitchers with 162+ IP in 2005 or 2006. The correlations for QERA (.55) and FIP (.67) are higher. I think I need to weight the events in order to get a really high correlation. But then it loses its simplicity and becomes another FIP or QERA stat.
ReplyDeleteIt does seem to be reasonably predictive though. As I said in the post, it was stable from 2005 to 2006 for 40 AL pitchers with 17 or more starts both years. Also, RPEPCT in 2005 has a .47 correlation with run average in 2006 for 40 pitchers with 17 or more starts in both years. The correlations with 2006 run average for other 2005 stats are: ERA .39, FIP .50 and QERA .40. 40 data points is not enough though so I'll look at this again when I get more years of data.
I had the exact same thoughts about weighting it.
ReplyDeleteI think the .50 is pretty significant in that it shows that a pitcher's ability to "pitch good" only accounts for half of what happens in terms of runs scoring.
Quick question -- where did the $126 million man rank? (I've been having fun with Bondo-Zito comparisons in recent days).
ReplyDeleteZito was #42.
ReplyDelete