League Leaders in ERA Estimators

by Lucas Apostoleris on Feb 1, 2011 2:01 PM EST in Research

Two stats I've been interested in recently are bbFIP and kwERA. These ERA estimators were created by Tom Tango as simple alternatives to FIP by using different components. bbFIP (batted ball fielding independent pitching) is used with strikeouts, walks, and batted ball types only (ground balls, fly balls, line drives, and pop ups). kwERA (strikeout-walk earned run average) is even simpler - its only inputs are strikeout and walk totals. Both metrics are slightly better than FIP at predicting future ERA performance (here is the link to the post at Tango's blog with the discussion on correlations in the comments section). I recently added these to my player pages with MLBAM data; in this post, I will show the leaders and trailers in 2010 for both metrics.

This is Tango's equation for bbFIP:

(((11*((BB+LD)-(K+PU)))+(3*(FB-GB)))/PA)+C

in which K is strikeouts, BB is unintentional walks and hit by pitches, GB is ground balls, FB is fly balls, LD is line drives, PU is pop-ups, PA is plate appearances, and C is the coefficient to align to the league average ERA.

Tango says, "A line drive is like a walk, an infield fly is like a strikeout, and the gap between an outfly and a groundball is about one-fourth the gap between BB and SO." I calculated the coefficient from Baseball Prospectus data, and I get 4.62 as the American League coefficient and 4.66 as the National League coefficient (scaling to 4.14 ERA in the AL and 4.03 ERA in the NL). Please take note that I'm using MLBAM data here, so batted ball classifications will be slightly different from the ones from BPro that I used to calculate the coefficients, so there should be a tiny difference in the values shown here than the ones calculated directly from BPro. I'll present the data in two tables: the first shows the 15 and leaders and trailers in bbFIP among pitchers with at least 200 but less than 500 opposing plate apperances, and the second one shows the data for pitchers with at least 500 opposing plate appearances.

Rank	Pitcher	Team	bbFIP
1	Joaquin Benoit	Rays	1.28
2	Billy Wagner	Braves	1.36
3	Hong-Chih Kuo	Dodgers	1.50
4	Rafael Betancourt	Rockies	1.71
5	Takashi Saito	Braves	1.89
6	Matt Thornton	White Sox	1.98
7	Sergio Romo	Giants	2.05
8	Brian Wilson	Giants	2.12
9	Ryan Madson	Phillies	2.19
10	J.J. Putz	White Sox	2.33
11	Stephen Strasburg	Nationals	2.33
12	Joel Hanrahan	Pirates	2.34
13	Edward Mujica	Padres	2.44
14	Neftali Feliz	Rangers	2.47
15	Luke Gregerson	Padres	2.48

1x	Dontrelle Willis	Tigers / Diamondbacks	6.13
2x	Oliver Perez	Mets	5.90
3x	Brad Lincoln	Pirates	5.78
4x	Casey Coleman	Cubs	5.78
5x	Dana Eveland	Blue Jays / Pirates	5.73
6x	David Huff	Indians	5.70
7x	Dusty Hughes	Royals	5.62
8x	Chris Tillman	Orioles	5.60
9x	Ian Snell	Mariners	5.52
10x	Lance Cormier	Rays	5.31
11x	Kyle Lohse	Cardinals	5.30
12x	Dustin Nippert	Rangers	5.28
13x	Aaron Laffey	Indians	5.26
14x	Jesse Litsch	Blue Jays	5.25
15x	Jeff Suppan	Brewers / Cardinals	5.24

Rank	Pitcher	Team	bbFIP
1	Jered Weaver	Angels	2.66
2	Hiroki Kuroda	Dodgers	2.81
3	Mat Latos	Padres	2.83
4	Francisco Liriano	Twins	2.99
5	Cliff Lee	Mariners / Rangers	3.00
6	Adam Wainwright	Cardinals	3.01
7	Cole Hamels	Phillies	3.06
8	Roy Halladay	Phillies	3.06
9	Colby Lewis	Rangers	3.09
10	Clayton Kershaw	Dodgers	3.09
11	CC Sabathia	Yankees	3.12
12	Roy Oswalt	Astros / Phillies	3.17
13	Felix Hernandez	Mariners	3.19
14	Josh Johnson	Marlins	3.21
15	Jon Lester	Red Sox	3.23

1x	Ryan Rowland-Smith	Mariners	5.76
2x	Scott Kazmir	Angels	5.28
3x	Kyle Kendrick	Phillies	5.15
4x	John Lannan	Nationals	5.15
5x	Dave Bush	Brewers	5.09
6x	Aaron Harang	Reds	5.04
7x	Livan Hernandez	Nationals	5.03
8x	Carlos Zambrano	Cubs	4.99
9x	Paul Maholm	Pirates	4.98
10x	Scott Feldman	Rangers	4.98
11x	Joe Saunders	Angels / Diamonbacks	4.90
12x	Brian Bannister	Royals	4.89
13x	Tom Gorzelanny	Cubs	4.87
14x	Rodrigo Lopez	Diamondbacks	4.87
15x	Craig Stammen	Nationals	4.84

Yes, the Phillies are going to be scary next year.

The formula I'm using for kwERA is C-(12*(K-BB)). The coefficients are 5.22 for the AL and 5.28 for the NL. You'll see a lot of the same names that were on the bbFIP leaderboard.

Rank	Pitcher	Team	kwERA
1	Rafael Betancourt	Rockies	1.27
2	Billy Wagner	Braves	1.58
3	Joaquin Benoit	Rays	1.65
4	Matt Thornton	White Sox	1.98
5	Stephen Strasburg	Nationals	2.01
6	Takashi Saito	Braves	2.32
7	Edward Mujica	Padres	2.33
8	Luke Gregerson	Padres	2.35
9	Carlos Marmol	Cubs	2.37
10	J.J. Putz	White Sox	2.40
11	Hong-Chih Kuo	Dodgers	2.42
12	Joel Hanrahan	Pirates	2.42
13	Ryan Madson	Phillies	2.49
14	Brian Wilson	Giants	2.51
15	John Axford	Brewers	2.57

1x	Dontrelle Willis	Tigers / Diamondbacks	5.74
2x	Dana Eveland	Blue Jays / Pirates	5.70
3x	Oliver Perez	Mets	5.70
4x	Jason Marquis	Nationals	5.37
5x	Lance Cormier	Rays	5.31
6x	Brad Thomas	Tigers	5.30
7x	Jesse Litsch	Blue Jays	5.28
8x	Sean Gallagher	Padres / Pirates	5.28
9x	Blaine Boyer	Diamondbacks	5.28
10x	Chris Tillman	Orioles	5.22
11x	Aaron Laffey	Indians	5.22
12x	Casey Coleman	Cubs	5.19
13x	Brian Moehler	Astros	5.18
14x	Jamey Wright	Indians / Mariners	5.18
15x	Ian Snell	Mariners	5.17

Rank	Pitcher	Team	kwERA
1	Cliff Lee	Mariners / Rangers	2.84
2	Jered Weaver	Angels	2.86
3	Mat Latos	Padres	3.01
4	Roy Halladay	Phillies	3.06
5	Josh Johnson	Marlins	3.10
6	Tim Lincecum	Giants	3.17
7	Cole Hamels	Phillies	3.22
8	Adam Wainwright	Cardinals	3.23
9	Francisco Liriano	Twins	3.24
10	Ricky Nolasco	Marlins	3.24
11	Brandon Morrow	Blue Jays	3.27
12	Dan Haren	Diamondbacks / Angels	3.29
13	Roy Oswalt	Astros / Phillies	3.32
14	Yovani Gallardo	Brewers	3.36
15	Felix Hernandez	Mariners	3.36

1x	Ryan Roland-Smith	Mariners	5.22
2x	Scott Kazmir	Angels	5.17
3x	Aaron Cook	Rockies	5.07
4x	Mitch Talbot	Indians	5.02
5x	Nick Blackburn	Twins	4.86
6x	Armando Galarraga	Tigers	4.86
7x	John Lannan	Nationals	4.83
8x	Paul Maholm	Pirates	4.81
9x	Brad Bergesen	Orioles	4.79
10x	Scott Feldman	Rangers	4.73
11x	Brian Bannister	Royals	4.73
12x	Kyle Kendrick	Phillies	4.72
13x	Vin Mazzaro	Athletics	4.69
14x	Mike Pelfrey	Mets	4.68
15x	Chris Volstad	Marlins	4.66

I will reiterate that the Phillies will be scary next year. Also, I must say that I didn't realize just how good Mat Latos is.

I find both metrics to be extremely intriguing. The thing that makes me more wary of bbFIP than of FIP or xFIP is that it uses line drives and pop ups as inputs; they don't normalize as quickly as ground ball and fly ball rates, so (as always), be careful about small sample sizes. I'd like to expand the data to the past three years so we can get a better feel for line drive and pop up skill. As for kwERA, I wish people would start using it, or even just (K-BB)/PA, instead of K/BB. Tango has shown that K-BB is superior to K/BB in terms of measuring run prevention.

For now, here is a spreadsheet with bbFIP and kwERA stats for all 635 pitchers in 2010 along with the total number of batters that each pitcher faced. For pitchers who pitched in both leagues, I'm weighting their averages by batters faced in each league.

The data in this post are courtesy of Joe Lefkowitz's PITCHf/x tool and Baseball Prospectus.

Tweet Comment 25 comments | 0 recs |

Do you like this story?

Comments

Ahhhh

there isn’t a metric in the world that makes Ollie Perez look good…

Beyond the Box Score

by Bill Petti on Feb 1, 2011 2:13 PM EST reply actions

I love how there are certain guys who can, to some extent, confirm the effectiveness of a new metric ...

by Lucas Apostoleris on Feb 1, 2011 2:18 PM EST up reply actions

Yes

I just wish that the team I root for would stop signing some of those players…

Writer at Beyond the Box Score and tortured Mets fan (is there any other kind?)

by Bill Petti on Feb 1, 2011 3:35 PM EST up reply actions

Well now it has.

by garik16 on Feb 1, 2011 3:36 PM EST up reply actions

I know

but it’s so hard to believe they’ve changed. Still doesn’t seem real.

Writer at Beyond the Box Score and tortured Mets fan (is there any other kind?)

by Bill Petti on Feb 1, 2011 3:48 PM EST up reply actions

Wins Below Replacement?

Baseball is my preferred sport. It should be yours, too.
I'm a columnist for Beyond the Box Score, an SB Nation blog.
Oh, I'm on Twitter, too.

by Satchel Price on Feb 1, 2011 3:33 PM EST via mobile up reply actions

I assume that the "constant" for kwERA is figured the same way as for FIP, right?

Making watching baseball as fun as doing your taxes.
My Twitter feed.

by Matt Klaassen on Feb 1, 2011 3:33 PM EST reply actions

Yes.

Basically, I applied the formula to the aggregated league data and found the difference between it and the league average ERA. The difference represents the coefficient.

by Lucas Apostoleris on Feb 1, 2011 3:44 PM EST up reply actions

Cool, I thought so

gives me an idea

Making watching baseball as fun as doing your taxes.
My Twitter feed.

by Matt Klaassen on Feb 2, 2011 9:27 AM EST up reply actions

Edward Mujica

The weird thing about Mujica was that I was always scared that he’d pitch poorly when he came in for the Padres last year. Maybe it was just because it seemed like he’d either strike someone out or give up a homerun. Even though his advanced stats looked good and his HR/FB rate is likely to regress I wasn’t sad to see the Padres trade him away. I probably should have been though. Having Latos should help reduce any suffering.

by Drakos on Feb 1, 2011 6:42 PM EST reply actions

I, too, love kwERA.

Mike Fast had a post on it at The Hardball Times and I used it to look at some college pitchers when I was trying to adjust college stats for a little bit. College starters in 2010 and relievers here.

Me Elsewhere:

Bless You Boys (Tigers Blog) / Beyond the Boxscore / My Michigan State (and Big Ten) Baseball Blog. / Western Michigan University Hoops / my Last.fm account.

by Mike Rogers on Feb 2, 2011 1:23 AM EST reply actions

Good stuff, Mike

I’m just curious … what was causing you to struggle with using FIP for college stats?

by Lucas Apostoleris on Feb 2, 2011 11:39 AM EST up reply actions

The coefficients aren't meant to be used in a college setting.

The run environment of college as a whole, and within the different conferences, varies so much from pro ball, that even adjusting just the 3.2 to the league/conference’s numbers, the weights for K’s and BB’s and HR’s were still likely wrong. It’s been a while since I’ve dealt with non-Big Ten baseball data, though. I’ve moved onto to Base Runs for pitching and hitting numbers.

Me Elsewhere:

Bless You Boys (Tigers Blog) / Beyond the Boxscore / My Michigan State (and Big Ten) Baseball Blog. / Western Michigan University Hoops / my Last.fm account.

by Mike Rogers on Feb 2, 2011 3:36 PM EST up reply actions

+1 to the K-BB instead of K/BB thing.

Not that Lee’s 1 BB/9 rate isn’t impressive or valuable. But he’s great because he can still strike out 8 per game and keep the in the park even though he’s pounding the strike zone.

Would be interesting to compare K-BB and WAR. Hmm, maybe I’ll try that.

@Sky_Kalkman

by Sky Kalkman on Feb 2, 2011 9:55 AM EST reply actions

Using rWAR (based on ERA, not FIP, which would probably be too similar to K,BB combinations)...

And limiting the pool to pitchers with 90+ IP in 2010 (to remove relievers and their leverage-adjusted WARs), there’s a .68 correlation between K-BB (total, not rate) and rWAR. Sounds pretty high.

@Sky_Kalkman

by Sky Kalkman on Feb 2, 2011 10:12 AM EST up reply actions

If instead of limiting IP, I limit aLI to be betwee .9 and 1.1, the correlation is .75.

@Sky_Kalkman

by Sky Kalkman on Feb 2, 2011 10:15 AM EST up reply actions

How about K/BB?

That might be tough considering you’re using counting stats, though.

by Lucas Apostoleris on Feb 2, 2011 10:30 AM EST up reply actions

Right. Maybe K/BB * IP?

Correl between IP and WAR: .66
Correl between SO/BB and WAR: .44
Correl between SO/BB*IP and WAR: .70 (small improvement over just IP)
Correl between (SO-BB)*IP and WAR: .77 (larger improvement over just IP, and also improvement over just SO-BB)
Correl between SO-BB and IP: .85

@Sky_Kalkman

by Sky Kalkman on Feb 2, 2011 10:45 AM EST up reply actions

Ooh, very cool. Thanks, Sky.

The last correlation has nothing to do with WAR, correct? Just K-BB and IP?

by Lucas Apostoleris on Feb 2, 2011 11:08 AM EST up reply actions

Correct.

IP is a surprisingly good measure of value (look at historical career leaders — many many HoFamers). Not just because more innings equals more value, but because you have to be good for your team to want you to pitch that much. And career-wise, you have to have been really good at your peak in order to be decent enough to pitch during your non-peak. (i.e. there’s a lot of selective sampling when you pick pitchers by IP.)

@Sky_Kalkman

by Sky Kalkman on Feb 2, 2011 11:15 AM EST up reply actions

Good point.

That’s why it’s also cool to look at the poor pitchers with a minimum of “x” IP. I think Adam did something like that here a month ago or so by looking at the fewest WAR compiled over a certain number of innings in a career.

by Lucas Apostoleris on Feb 2, 2011 11:25 AM EST up reply actions