Friday, November 25, 2011

Combining the DIPS Metrics into One

In an earlier post, I used the Based Runs Saved Above Average (BsRAA) statistic to estimate runs saved by pitchers based on statistical components - hits, walks, homers, total bases and outs.  As the above post describes, BsRAA removes timing of events out of the pitching equation.

For example, ERA rewards a pitcher for pitching well with runners on base and penalizes a pitcher for clustering lots of base runners in one inning.  BsRAA assumes that a pitcher does not have a lot of control over such sequencing of events and thus ignores them.  furthermore, unlike ERA, a starter's BsRAA is independent of how relief pitchers perform once he leaves the game.

One thing BsRAA does include which pitchers don't control by themselves is hits allowed.  In an article entitled “Pitchers and Defense: How much control do hurlers have?” published at
Baseball Prospectus in 2001, researcher Voros McCracken presented his Defense Independent
Pitching Statistics (DIPS) theory. The results of his study suggested that there is surprisingly
little difference among pitchers in their ability to prevent hits on balls put into the field of play and that hits allowed are not very meaningful in the evaluation of a pitcher.

Specifically, McCracken revealed that there is very little correlation between pitcher quality and the relative frequency of hits allowed on balls in play. He also showed that pitchers have much less control over hits allowed than they do over defense independent events like walks and strikeouts.

McCracken’s somewhat counter-intuitive conclusions were met with much skepticism, even in the sabermetric community. Tom Tippett, currently a statistician for the Boston Red Sox,  did an extensive study in 2003 which revealed that some pitchers such as Tom Glavine and Jamie Moyer have been consistently above average in terms of preventing hits on balls in play. Tippett also found that knuckleball pitchers, in general, had been particularly good at limiting hits. Two examples are Charlie Hough and Phil Niekro.

Mitchel Lichtman further challenged McCracken’s conclusions in a study of the impact of batted ball types on hits on balls in play in 2004. Lichtman found that pitchers had a good deal of control over how many ground balls and fly balls (in both the infield and outfield) they allowed and that ground balls were more likely to be hits than fly balls.

Analysts continue to explore the DIPS theory today. The most recent effort is being made by Mike Fast at Baseball Prospectus who is looking at differences in abilities of pitchers to avoid quality contact on batted balls

While there are still questions about how much control pitchers have over hits allowed, it is useful to have ways of estimating how well they perform on items they do control.  The most common DIPS statistic is Fielding Independent Pitching ERA (FIP) created by Tom Tango.  It estimates pitcher ERA based on BB, HBP, SO and HR, four events which are largely unaffected by fielders.

Pitchers have also shown varying degrees of control over the percentages of batted balls that result in ground balls (GB%), fly balls (FB%), pop ups (PO%) and line drives (LD%) allowed.  Thus, other statistics have been developed to incorporate batted ball data.  Because pitchers have more control over the number of fly balls allowed than home runs, David Studenmund  developed Expected FIP (xFIP)  which is based on BB, HBP, SO, fly balls allowed and the MLB HR/FB rate of about 11% .

Two other measures which include batted ball data are True Earned Run Average (tERA) created by Matthew Curruth and Graham MacAree and Skill-Interactive ERA (SIERA) designed by Matt Swartz and Eric Seidman. The tERA statistic includes BB, HBP, SO, HR, GO%, FB%, PO%, LD%.  The SIERA metric uses BB, SO, GB%, FB% and PO%.  

None of above metrics - FIP, xFIP, tERA or SIERA - should be used to measure a pitcher's overall performance, but they are useful in evaluating how good a pitcher is at things he most controls independent of teammates.  Which DIPS measure is best depends on how complicated you want to get and how much you trust batted ball data.

I don't have a strong preference among the DIPS measures, so I'll compute the average of the four and call it Weighted DIPS  (WDIP).  Table 1 below shows WDIP for the top starters in the American League in 2011.  Not surprisingly, AL Cy Young and MVP winner Justin Verlander was the leader at 3.05.  Doug Fister was eighth at 3.43 pitching for the Mariners and Tigers.

The final column incudes ERA for comparison with WDIP.  Verlander's WDIP was 0.65 higher than his ERA (2.40).  One reason was his success at pitching with runners on base as evidenced by his 80.3 Left On Base Percentage.  Another explanation was his very low Batting Average on Balls in Play (.236).  Reasons for this could be strong defensive support (not likely with the Tigers fielders), ability to induce weak contact on batted balls and perhaps the absence of lucky hits (bloopers and bleeders).   

Table 1: AL Weighted DIPS Leaders, 2011

Player
Team
IP
FIP
XFIP
TERA
SIERA
WDIP
ERA
Justin Verlander
DET
251.0
2.99
3.12
3.09
2.99
3.05
2.40
CC Sabathia
NYY
237.1
2.88
3.02
3.75
3.14
3.20
3.00
Dan Haren
LAA
238.1
2.98
3.29
3.21
3.34
3.21
3.17
Felix Hernandez
SEA
233.2
3.13
3.15
3.34
3.22
3.21
3.47
Brandon McCarthy
OAK
170.2
2.86
3.30
3.28
3.49
3.23
3.32
David Price
TBR
224.1
3.32
3.32
3.39
3.27
3.33
3.49
James Shields
TBR
249.1
3.42
3.25
3.47
3.29
3.36
2.82
Doug Fister
AL
216.1
3.02
3.61
3.40
3.67
3.43
2.83
Michael Pineda
SEA
171.0
3.42
3.53
3.42
3.36
3.43
3.74
C.J. Wilson
TEX
223.1
3.24
3.41
3.75
3.44
3.46
2.94

Data source: FanGraphs.com

WDIP is a rate statistic which does not credit pitchers for innings pitched.  We can convert it to a runs saved measure by adding innings pitched, earned runs and average ERA to the equation.  For example, Verlander had a 3.05 WDIP in 251 innings in 2011.  The league average ERA for AL starters was 4.23, so you would expect the average pitcher to allow 251*4.23/9 = 118 earned runs.  Verlander allowed 67, so he had 118 - 67 = 51 DIP Runs Saved (DIPRS).


Table two lists the AL pitchers with the most DIPRS in 2011.  As expected, Verlander was the leader with 51 followed by Angels ace Jered Weaver at 48.  Fister came in fourth at 34.

Table 2: AL DIPs Runs Saved Leaders, 2011

Player
Team
IP
WDIP
DIPRS
Justin Verlander
DET
251.0
3.05
51
Jered Weaver
LAA
235.2
3.47
48
James Shields
TBR
249.1
3.36
39
Doug Fister
AL
216.1
3.43
34
Ricky Romero
TOR
225.0
3.97
33
CC Sabathia
NYY
237.1
3.20
32
C.J. Wilson
TEX
223.1
3.46
32
Josh Beckett
BOS
193.0
3.60
29
Dan Haren
LAA
238.1
3.21
28
Jeremy Hellickson
TBR
189.0
4.61
27

Data source: FanGraphs.com

No comments:

Post a Comment

Sabermetrics Book

Sabermetrics Book
One of Baseball America's top ten books of 2010

Blog Archive

Subscribe

501 Baseball Books

501 Baseball Books
Recommended by Tiger Tales

Stat Counter

Site Meter