Re-WERCing Things

Linear Weights guru Tom Tango likes to use wOBA against for pitchers as well as hitters.  However, he told me yesterday that Base Runs is preferable to trying to convert wOBA to runs as I did in my last post.  So, I'm going to do some leader boards according to Base Runs today.  It turns out they aren't radically different from the linear weights leaders presented my previous post, although linear weights seemed to inflate the value of star pitchers (runs saved above average) a little bit

The Base Runs measure was created by David Smythe in the early 1990s.  It is based on the idea that we can estimate team runs scored if we know the number of base runners, total bases, home runs and the typical score rate (the score rate is the percentage of base runners that score on average.  Base Runs also works well for individual pitchers.  The complete formula can be found here.

Justin Verlander has 39 Base Runs Against in 151 innings so far this year.  This means that he should have allowed an estimated 39 runs based on the number of base runners, total bases and home runs he has allowed.  He has allowed 39 actual runs, so runs are scoring against him at the exact rate you would expect.

Verlander has allowed 33 Base Runs Above Average (RAA) which means that he has saved the Tigers an estimated 33 runs compared to the average pitcher in the same number of innings.  Table 1 shows that he is tied for first in the American League with Angels ace Jered Weaver.


Table 1 - AL Base Runs Allowed Above Average Leaders as of July 10

Player	Team	IP	Base Runs Against	RAA
Jered Weaver	LAA	140.1	33	33
Justin Verlander	DET	151.0	39	33
Josh Beckett	BOS	111.0	27	25
James Shields	TBR	142.2	44	24
Dan Haren	LAA	134.1	40	24
CC Sabathia*	NYY	145.2	47	22
Felix Hernandez	SEA	144.0	49	19
Michael Pineda	SEA	113.0	35	18
Philip Humber	CHW	107.1	34	17
Alexi Ogando	TEX	104.2	33	16

Table 2 shows that Verlander has allowed 2.32 Base Runs per nine innings.  About 93% of runs are earned, so multiply this result by .93. to put it on the same scale as ERA. The final result is a weighted component ERA.  Although, I am no longer using linear weights here, I'll keep calling it WERC because others have said the like the name. It's really not a novel idea though.  U.S. Patriot of Walk Like a Saber has been using Base Runs to evaluate pitchers for a while but prefers to not convert to the ERA scale.

Getting back to the example, Verlander has a 2.16 WERC.  Since his Base Runs Allowed is the same as his actual runs allowed, this is almost the same as his real ERA.  He is third in the league behind Weaver (1.98) and Josh Beckett of the Red Sox (2.06). 


Table 2: AL WERC Leaders as of July 10

Player	Team	IP	Base Runs/9 IP	WERC
Jered Weaver	LAA	140.1	2.13	1.98
Josh Beckett	BOS	111.0	2.21	2.06
Justin Verlander	DET	151.0	2.32	2.16
Dan Haren	LAA	134.1	2.68	2.49
James Shields	TBR	142.2	2.77	2.58
Michael Pineda	SEA	113.0	2.83	2.63
Philip Humber	CHW	107.1	2.87	2.67
Alexi Ogando	TEX	104.2	2.89	2.68
CC Sabathia*	NYY	145.2	2.92	2.72
Felix Hernandez	SEA	144.0	3.09	2.87

The raw data used to create the statistics in this post was extracted from Baseball-Reference.

Scherzer, Sanchez Topped AL in Run Prevention

Tigers right hander Max Scherzer is probably going to win the American League Cy Young Award by a wide margin based mostly on traditional statistics.  He led the league in wins with a gaudy 21-3 record, finished second with 240 strikeouts and fifth in ERA at 2.90.  However, there were several other starting pitchers who had excellent seasons and should also get consideration for the award including teammate Anibal Sanchez, Rangers ace Yu Darvish, Hishashi Iwakuma of the Mariners and White Sox southpaw Chris Sale among others.

There is no surefire way to determine the best pitcher in the league, but a pitchers job is to prevent runs.  So, it's useful to estimate how many runs pitchers saved their teams compared to an average pitcher.  As the regular season wound down, I explored four different ways to do this:   

• Pitching Runs -  Runs Saved Above Average based on innings and runs allowed. 

• Adjusted Pitching Runs - Pitching Runs minus estimated Runs allowed by fielders. 

• Base Runs -  Runs Saved Above Average based on batters faced and hits, walks, total bases and home runs allowed.

• FIP Runs - Runs Saved Above Average based on innings, bases on balls, hit batsmen and home runs allowed and strikeouts.

Scherzer had 29 Pitching Runs in 2013 which indicates that he saved the Tigers an estimated 29 runs compared to the average pitcher in the same number of innings.  This placed him fifth in the league behind   Iwakuma (36),  Darvish (32), Sanchez (31) and Athletics right hander Bartolo Colon (31).  A statistic based on pure runs allowed is a good place to start, but it does not take into consideration things such as team defense and ballparks.

If we attempt to take team defense out of the equation, then Scherzer looks better since Detroit had one of the three worst defenses according to Defensive Runs Saved and Total Zone in the AL along with the Mariners and White Sox.  According to Adjusted Pitching Runs which considers defense as well as ballpark, Scherzer finished with 37 Adjusted Pitching Runs which was second to Sanchez (38). The biggest limitations of this metric arer the uncertainty of the fielding measures and the assumption that all pitchers are affected by team defense in the same way. 

Even if you trust the fielding component of the Adjusting Pitching Runs calculation, another issue is that a pitcher has no control over what happens after he leaves the game   If he departs with a man on first with two outs and the relief pitcher allows a run-scoring double, the starting pitcher is charged with the run. In other words, a pitcher’s ERA is dependent not only on the quality of his innings but also on the quality of the innings of his relievers.

Another potential concern regarding Pitching Runs and Adjusted Pitching Runs is the timing of hits, walks and extra base hits. For example, if a pitcher pitches nine innings and gives up nine hits with each hit coming in a different inning, he will almost surely allow fewer runs than if he surrenders all the hits in one
inning. If a pitcher frequently allows a lot of baserunners and extra base hits, he might get away with a relatively low ERA one year but it wouldn't necessarily be based on skill.

A related issue to the distribution of base runners is sequencing of events. Let’s say a pitcher
allows the following sequence of events in an inning:

1. Ground out
2. Single
3. Single
4. Homer
5. Strikeout
6. Fly out

In this case, he would be charged with three runs allowed for the inning. Now, suppose that
Pitcher B has a slightly different sequence of events in another inning:

1. Ground out
2. Homer
3. Single
4. Single
5. Strikeout
6. Fly out

In this case, the pitcher is charged with one run. Both pitchers surrendered a homer and two
singles but Pitcher B allowed two fewer runs just because the sequence of hits was different.

While pitchers vary in their ability to prevent baserunners from scoring, research by Ron Shandler – author of The Baseball Forecaster and publisher of BaseballHQ.com – suggests that this has more to do with overall pitcher quality than clutch pitching ability. In other words, most pitchers who consistently strand runners do so primarily because they get strikeouts and limit base runners in all situations, not because they have a lot of control over clustering of base runners or sequencing of events.  In other words, much of the bunching and sequencing seems to based on luck to some extent. 

In order to remove, clustering of base runners and sequencing of events from the equation, we can used a component-based statistic.  One such measure is the Base Runs statistic created by David Smythe in the early 1990s.  It is based on the idea that we can estimate team runs scored if we know the number of base runners, total bases, home runs and the typical score rate (the score rate is the percentage of base runners that score on average).  Base Runs also works well for individual pitchers.  The complete formula can be found here.  Scherzer had a comfortable lead with 37 Base Runs which was 10 runs better than Darvish (27) the runner up.

A criticism of Base Runs for evaluation of pitchers is that it includes hits on balls in play which is the responsibility of fielders at least as much as pitchers.  Thus, many analysts prefer to use FIP (translated in to FIP Runs here) which only considers events that a pitcher essentially controls - walks, hit batsmen and home runs allowed and strikeouts.  Scherzer had 31 FIP Runs which was second to Sanchez at 33. 

Table 1 below lists all four statistics discussed above - Pitching Runs, Adjusted Pitching Runs, Base Runs and FIP Runs - side by side and also the average of the four for the top fifteen pitchers.  Scherzer's average across the measures was 33 which was the best in the league.  He was followed by Sanchez (31) and Darvish (29).

So, based on this aggregate measure, Scherzer appears to be the deserving winner of the award he will likely soon reap.

Table 1: American League Runs Above Average Leaders, 2013

Pitcher	Team	IP	Pitching Runs	Adjusted Pitching Runs	Base Runs	FIP Runs	Average
Max Scherzer	DET	214.1	29	37	37	31	33
Anibal Sanchez	DET	182.0	31	38	24	33	31
Yu Darvish	TEX	209.2	32	36	27	22	29
Hisashi Iwakuma	SEA	219.2	36	37	26	7	26
Chris Sale*	CHW	214.1	21	33	26	23	26
Felix Hernandez	SEA	204.1	23	24	23	27	24
Bartolo Colon	OAK	190.1	31	32	18	14	24
Justin Masterson	CLE	193.0	17	18	26	11	18
James Shields	KCR	228.2	27	14	16	8	16
Hiroki Kuroda	NYY	201.1	17	21	13	12	16
Justin Verlander	DET	218.1	10	19	8	19	14
David Price*	TBR	186.2	11	7	19	17	13
Doug Fister	DET	208.2	8	17	9	18	13
Jose Quintana*	CHW	200.0	12	24	8	8	13
Derek Holland*	TEX	213.0	12	14	6	18	12

  Data source: Baseball-Reference.com