Wednesday, July 17, 2013

Ranking Pitchers By Fielding Independent ERA Adjusted For The League Average And Park Effects

I am not sure how much sense this makes and someone else may have done this. There may be problems with this approach that I have not thought of (and there are some that did occurr to me that maybe I could discuss in the future).

Here is how it works: I used Fielding Independent ERA from Fangraphs. But first I used RSAA or "Runs Saved Above Average" from the Lee Sinins Complete Baseball Encyclopedia. It adjusts for the league average and park effects.

Take Pedro Martinez, for example. He had 2,827.33 IP and an RSAA of 496. So he saved about 1.58 runs per 9 IP. Suppose we are in a league that has an average of 4 runs per game. It means he would allow about 2.42 runs per game.

But, according to Fangraphs, Martinez had an FIP ERA of 2.91, .02 lower than his actual ERA. So I subtracted .02 from 2.42 to get 2.40. Below is the top 25 pitchers with 2500+ IP from 1876-2011.

Pedro Martinez2.40
Roger Clemens2.63
Randy Johnson2.75
Lefty Grove2.77
Curt Schilling2.82
Roy Halladay2.86
Rube Waddell2.99
Mike Mussina3.07
Bret Saberhagen3.08
John Smoltz3.10
Greg Maddux3.11
Bob Gibson3.17
Dazzy Vance3.18
Cy Young3.20
Hal Newhouser3.20
Kevin Brown3.21
Walter Johnson3.21
Andy Pettitte3.22
Bert Blyleven3.26
Kevin Appier3.29
Kid Nichols3.31
Dizzy Trout3.31
Dennis Eckersley3.33
Christy Mathewson3.34
Rick Reuschel3.34
 
If I get time tomorrow, I will create a link with all 238 pitchers. Koufax is at about 3.08. But he only had  2325 IP. Catfish Hunter was 229th at 4.25.

I also used relative runs allowed. Martinez, for example allowed 1006 runs in his career. Since he saved 496, that means the average pitcher would have allowed 1502. Since 1006/1505 is about .67, I gave him about 2.68 runs allowed per game (since .67*4 = 2.68). But then I lowered it by .02. To get 2.66. Here are the top 25 using the relative method

Pedro Martinez2.66
Roger Clemens2.85
Curt Schilling2.94
Randy Johnson2.95
Rube Waddell2.99
Roy Halladay3.06
Lefty Grove3.12
Bret Saberhagen3.18
John Smoltz3.18
Walter Johnson3.18
Bob Gibson3.19
Mike Mussina3.20
Greg Maddux3.23
Ed Walsh3.25
Dazzy Vance3.27
Christy Mathewson3.30
Bert Blyleven3.30
Hal Newhouser3.31
Kevin Brown3.32
Andy Pettitte3.34
Eddie Plank3.37
Cy Young3.37
Rick Reuschel3.37
Dennis Eckersley3.38
Chief Bender3.38

7 comments:

Smada said...

Wanted to let you know I've got you on RSS and haven't had a post that I've disliked. Keep it up

Cyril Morong said...

Great, thanks. Glad you like it

David said...

For my money, I probably would have used 4.5 R/G. It strikes me as a little more historically accurate (though I could be wrong). Plus, the whole half-run-per-inning thing is pretty nice if you use 4.5 instead of 4.0.

Second of all, while RSAA does adjust for league and park, does it adjust for era? In other words, saving a half-run in 2001 or 1930 would be a big deal, when the teams combine for about 11 R/G. But not NEARLY as big a deal as saving a half-run in 1968, when the teams were combining for something like a 7 R/G environment. So I would think that needs to be accounted for. If, for example, Pedro's run environment in his career was something like a 10 R/G environment (that should be fairly close), and he saved 496 runs, that's about 50 "games" worth of runs. If he had pitched in a 9 R/G environment, but STILL had 496 RSAA, that would be over 55 "games" worth of runs. It seems to me that this should be adjusted for somehow since one run in one era is not necessarily equal to one run in a different era. Simply hacking off the runs as if they were equal gives a false impression, I would think. Basically, I wouldn't be surprised if it bumped up guys like Saberhagen, Waddell, and Gibson. Just food for thought.

Cyril Morong said...

Thanks for reading and commenting.

Your raise some good points. I don't have any ideas on how to make the adjustments that would take into account what you mention.

I did once do a study based on runs per win. Here is the link

http://baseballanalysts.com/archives/2005/07/ranking_the_bes_1.php

But I did not use FIP ERA in that case. Just RSAA

I thought by using the % of runs saved, it might be a bit better. Because I did think that it would be easier to go below average in a high run environment.

The RSAA does take into account the league average. But as I said, it might be easier to be 1 run below average over 9 innings when the average is 5 runs per game than 3 runs per game. So that is why I used the % method. That would make saving 1 run per 9 IP in a 5 run environment equal to saving .6 in a 3 run environment.

I could adjust the RSAA the way I do it here and then calculate a runs per win and then a win total. I was also thinking of converting this to WAR

David said...

Yeah, it gets pretty tough unless you convert runs to wins. Then you start getting somewhere. But, of course, that's not what you're looking at: you're looking at rate stuff. I guess I'd recommend using the percentage method (or maybe a standard deviation method). Of course, I would use the relative percentage of FIP to ERA, rather than a straight addition/subtraction, again to account for the runs:wins ratio (but frankly, it's probably not that big of a deal). Still, this is a good idea. Good post, and thanks for indulging my thoughts!

Cyril Morong said...

If I do wins based on runs saved and the context (run environment) and the runs saved also reflect FIP ERA, here are the top 25. That is using the formula which says "it takes 10 times the square root of the number of runs scored per inning by both teams"

Roger Clemens 72.53
Cy Young 63.23
Walter Johnson 55.85
Randy Johnson 55.17
Lefty Grove 50.08
Greg Maddux 49.35
Pedro Martinez 48.73
Curt Schilling 42.50
Bert Blyleven 41.98
Christy Mathewson 38.27
Bob Gibson 37.57
Grover C Alexander 37.29
Gaylord Perry 36.27
Mike Mussina 35.92
Nolan Ryan 35.79
Rube Waddell 35.31
John Smoltz 35.01
Kid Nichols 34.54
Steve Carlton 33.55
Tom Seaver 33.08
Eddie Plank 32.32
Roy Halladay 31.09
Ferguson Jenkins 30.98
Kevin Brown 28.31
Jack Quinn 28.25

Cyril Morong said...

If I calculate WAR using a .360 pct for a replacement pitcher and the Pythagorean formula, here is the top 25

Cy Young 183.44
Walter Johnson 166.15
Roger Clemens 165.58
Greg Maddux 136.65
Randy Johnson 131.86
Bert Blyleven 129.77
Nolan Ryan 128.92
Gaylord Perry 128.45
Grover C Alexander 126.54
Christy Mathewson 124.95
Steve Carlton 122.71
Tom Seaver 117.03
Lefty Grove 114.67
Eddie Plank 112.47
Bob Gibson 108.52
Ferguson Jenkins 108.18
Curt Schilling 104.90
Pedro Martinez 104.77
Kid Nichols 103.52
Don Sutton 102.22
Phil Niekro 100.59
Pud Galvin 99.06
Mike Mussina 98.97
John Smoltz 97.69
Tim Keefe 96.69