/cdn.vox-cdn.com/photo_images/3705655/121308050.jpg)
A few weeks ago, I wrote a post about whether or not Zack Greinke's focus on FIP was actually hurting him as a pitcher, especially in terms of BABIP. The results were fairly inconclusive, mainly because of small sample size (only about three seasons) and the lackluster defense that has been played behind Greinke in his career.
Greinke's high BABIP's despite his ability to combine a lot of strikeouts, while allowing few walks and home runs, was enough to inspire Tom Tango, the inventor of FIP, to write a response to my original post, on insidethebook.com. Tango brought up two interesting points, in his response. His first point was about "last generation's version of Greinke" (in a sense, of course), Curt Schilling.
Schilling sports an incredible career K/BB rate of 4.38 and gave up less home runs than the average pitcher over his career (0.96 HR/9, despite playing in some home run-friendly parks), yet his career BABIP is .293. In the 11 seasons in which Schilling was a qualified starter (162+ IP) he lead the league in K/BB-rate three times (2001-02,2006), finished in the top-5 six times, and finished seventh one other time; thus, he was in the league's top-10 in seven out of the 11 seasons.
Yet, Schilling finished with a BABIP that was among the top-20 highest in baseball four times, three of which came in the seasons that he lead the league in K/BB. This is the interesting aspect of Schilling's career that Tango refers to, " there is a modest relationship between a low-BABIP and a high-K rate. So, Curt really bucks the trend here." Schilling seems to display all of the qualities that I was attempting to claim Greinke had, but are these two pitchers really that similar? I listed their career numbers below:
ERA |
FIP |
K/BB |
HR/9 |
BABIP |
|
Schilling |
3.46 |
3.23 |
4.38 |
0.96 |
0.293 |
Greinke |
3.8 |
3.46 |
3.52 |
0.90 |
0.309 |
These numbers are pretty similar, but I would I wouldn't go as far as to say they are perfect comps for each other. However, what should be considered is that Greinke is just 28 years old and only became a good pitcher in terms of K/BB in his age-25 season; while Schilling didn't start having high K/BB numbers until his age-28 season, and really became great at that statistic from age 33 until he retired, at age 40.
So Greinke still has a ton of time to put up the K/BB numbers that Schilling did in his career. Even though Greinke has more time there are still similarities in their numbers once the pitchers became "good at K/BB". Since Greinke turned 25 he has a K/BB-rate of 4.16 with a BABIP of .310, while Schilling posted a K/BB of 5.27 and BABIP of .297 from age 28 to the end of his career. The difference in BABIP has a lot to do with team defense, but also could be caused by Schilling's higher home run rate, 1.03 HR/9 to Greinke's 0.66, as more balls were put into play against Greinke.
The second comment Tango made in his response was even more interesting than the complexity of Schilling's career, and Greinke's possible future. Tango points out that "pitchers who have a lot of BB and a lot of HR end up with a low BABIP because one you remove the "easy" hits (HR)." Schilling and Greinke are the opposite of the type of low-BABIP pitchers that Tango refers to. On Greinke's SABR page he states that, "I try to get ahead in the count without leaving it run down the middle in a person's power zone. That helps me not walk guys," while Tango points out that Schilling "likely pitches closer to the center than an average pitcher."
Thus it seems that Greinke and Schilling are pitchers who are in the strike zone more often than most with good stuff. Does this mean they pitch more "hittable" balls than the average guy? By hittable I mean pitches that are more prone to ending in a base hit in play (heavy emphasis on in play), because these pitchers have good enough stuff to strikeout a lot of batters and keep the ball in the ballpark. To see if there was any weight to this theory I took a sample of qualified starters who had seriously above-average K/BB-rates (4.50 or above) coupled with a HR/9 below league average, over the last 15 MLB seasons.
**Note-- The list is fairly long, so I would suggest just scrolling past it and getting back to the analysis. I still listed it here for those who are interested in seeing all of the names and numbers, as well as, allowing readers to refer back to the table as I move through the rest of my analysis.
Pitcher Season K/BB HR/9 BABIP FiP (Rank) ERA (Rank) fWAR (Rank) Greg Maddux 1997 8.85 0.35 0.280 2.43 (3rd) 2.20 (3rd) 8.2 (4th) Curt Schilling 1997 5.5 0.88 0.304 2.62 (4th) 2.97 (12th) 8.6 (3rd) John Burkett 1997 4.63 0.95 0.346 3.55 (17th) 4.56 (65th) 5.0 (14th) Pedro Martinez 1997 4.55 0.60 0.258 2.39 (2nd) 1.90 (1st) 8.8 (2nd) Kevin Brown 1998 5.24 0.28 0.306 2.25 (1st) 2.39 (2nd) 9.2 (1st) Curt Schilling 1998 4.92 0.77 0.305 2.77 (4th) 3.25 (11th) 8.6 (3rd) Greg Maddux 1998 4.53 0.47 0.262 2.81 (5th) 2.22 (1st) 7.6 (5th) Pedro Martinez 1999 8.46 0.38 0.323 1.33 (1st) 2.07 (1st) 12.0 (1st) Shane Reynolds 1999 5.32 0.89 0.322 3.22 (5th) 3.85 (19th) 6.2 (4th) Randy Johnson 1999 5.2 0.99 0.292 2.76 (2nd) 2.48 (2nd) 9.8 (2nd) Pedro Martinez 2000 8.88 0.71 0.236 2.17 (1st) 1.74 (1st) 10.1 (1st) David Wells 2000 5.35 0.90 0.327 3.50 (5th) 4.11 (27th) 6.8 (4th) Kevin Brown 2000 4.6 0.82 0.255 3.17 (3rd) 2.58 (2nd) 6.2 (6th) Randy Johnson 2000 4.57 0.83 0.326 2.53 (2nd) 2.64 (3rd) 9.7 (2nd) Mike Mussina 2000 4.57 1.06 0.297 3.52 (7th) 3.79 (15th) 6.4 (5th) Greg Maddux 2000 4.52 0.69 0.274 3.23 (4th) 3.00 (4th) 7.2 (3rd) Greg Maddux 2001 6.41 0.77 0.286 3.12 (6th) 3.05 (4th) 6.3 (5th) Brad Radke 2001 5.27 0.96 0.292 3.70 (18th) 3.94 (30th) 5.3 (11th) Randy Johnson 2001 5.24 0.68 0.315 2.22(1st) 2.56 (1st) 10.0 (1st) Mike Mussina 2001 5.1 0.79 0.289 2.92 (2nd) 3.15 (7th) 7.1 (3rd) Javy Vazquez 2001 4.73 0.97 0.279 3.21 (7th) 3.42 (17th) 6.4 (4th) Curt Schilling 2002 9.58 1.01 0.297 2.35 (2nd) 3.14 (14th) 9.7 (1st) Pedro Martinez 2002 5.98 0.59 0.273 2.24 (1st) 2.26 (1st) 8.3 (3rd) Randy Johnson 2002 4.7 0.90 0.289 2.66 (3rd) 2.32 (2nd) 8.7 (2nd) Roy Halladay 2003 6.38 0.88 0.284 3.23 (8th) 3.25 (15th) 8.0 (1st) Curt Schilling 2003 6.06 0.91 0.297 2.66 (4th) 2.95 (8th) 5.9 (10th) David Wells 2003 5.05 1.01 0.297 3.97 (32nd) 4.21 (54th) 3.9 (26th) Mark Prior 2003 4.9 0.64 0.309 2.47 (2nd) 2.43 (4th) 7.6 (3rd) Mike Mussina 2003 4.88 0.88 0.287 3.09 (7th) 3.40 (19th) 6.4 (6th) Jason Schmidt 2003 4.52 0.61 0.253 2.64 (3rd) 2.34 (2nd) 6.7 (5th) Ben Sheets 2004 8.25 0.95 0.288 2.65 (2nd) 2.70 (4th) 8.0 (2nd) Randy Johnson 2004 6.59 0.66 0.264 2.30 (1st) 2.60 (2nd) 9.9 (1st) Curt Schilling 2004 5.8 0.91 0.284 3.11 (5th) 3.26 (12th) 7.3 (4th) Jon Lieber 2004 5.67 1.02 0.323 3.71 (18th) 4.33 (50th) 4.0 (24th) Brad Radke 2004 5.5 0.94 0.293 3.55 (13th) 3.48 (16th) 5.7 (9th) David Wells 2004 5.05 1.06 0.274 3.88 (24th) 3.73 (27th) 2.7 (51st) Johan Santana 2004 4.91 0.95 0.250 2.92 (3rd) 2.61 (3rd) 7.7 (3rd) Johan Santana 2005 5.29 0.85 0.262 2.80 (1st) 2.87 (7th) 7.6 (1st) David Wells 2005 5.1 1.03 0.320 3.83 (29th) 4.45 (66th) 4.1 (22nd) Johan Santana 2006 5.21 0.92 0.269 3.04 (1st) 2.77 (1st) 7.3 (1st) Mike Mussina 2006 4.91 1.00 0.284 3.46 (9th) 3.51 (11th) 5.2 (11th) CC Sabathia 2007 5.65 0.75 0.311 3.14 (3rd) 3.21 (11th) 7.1 (1st) Josh Beckett 2007 4.85 0.76 0.304 3.08 (2nd) 3.27 (12th) 6.5 (3rd) Roy Halladay 2008 5.28 0.66 0.284 3.05 (5th) 2.81 (5th) 7.3 (3rd) Dan Haren 2008 5.15 0.79 0.303 3.01 (4th) 3.33 (17th) 6.5 (5th) Josh Beckett 2008 5.06 0.93 0.315 3.24 (6th) 4.03 (49th) 5.1 (13th) Cliff Lee 2008 5 0.48 0.301 2.83 (2nd) 2.54 (2nd) 7.2 (4th) Mike Mussina 2008 4.85 0.76 0.321 3.32 (10th) 3.37 (18th) 5.3 (9th) Ervin Santana 2008 4.55 0.95 0.289 3.30 (9th) 3.49 (23rd) 5.8 (7th) Roy Halladay 2009 5.94 0.83 0.306 3.06 (6th) 2.79 (7th) 7.4 (4th) Javy Vazquez 2009 5.41 0.82 0.283 2.77 (3rd) 2.87 (9th) 6.5 (7th) Zack Greinke 2009 4.75 0.43 0.303 2.33 (1st) 2.16 (1st) 9.3 (1st) Cliff Lee 2010 10.28 0.68 0.287 2.58 (2nd) 3.18 (21st) 7.2 (1st) Roy Halladay 2010 7.3 0.86 0.290 3.01 (7th) 2.44 (5th) 6.6 (2nd) Roy Halladay 2011 6.29 0.39 0.298 2.20 (1st) 2.35 (2nd) 8.2 (1st) Dan Haren 2011 5.82 0.76 0.272 2.99 (10th) 3.19 (21st) 6.3 (6th) Cliff Lee 2011 5.67 0.70 0.291 2.60 (3rd) 2.40 (3rd) 6.7 (5th) Brandon McCarthy 2011 4.92 0.58 0.296 2.86 (5th) 3.32 (27th) 4.7 (20th) Clayton Kershaw 2011 4.59 0.58 0.269 2.47 (2nd) 2.28 (1st) 6.8 (4th)
There were 59 individual pitcher seasons that qualified for this sample. About four pitchers per season, over the last 15 years, have been good enough in terms of K/BB and HR/9 to be a part of this list. 30 out of the 59 (50.8%) pitchers in the sample had a BABIP at league average or higher. When the minimum K/BB-rate is moved up from 4.5 to 5.0, the sample gets cut down to 38 seasons; 20 of those 38 pitchers' seasons resulted in a BABIP at or above the league average (52.6%).
There's only a minor uptick in pitchers with higher BABIP's when the minimum parameter for K/BB is set higher, and that percentage increase is an insignificant amount. At first glance these percentages seem to make a fair amount of sense. One would expect that 50% of of pitchers in any given (random) large-ish sample would have a BABIP below league average, while about 50% would scatter above the league average. But what is critical about this sample is that it is the exact opposite of random.
The given structure of parameters caused this group of pitchers to be handpicked from all the starters in baseball. These pitchers aren't random, but instead are the one's who are adept at striking out an above average amount of hitters, while giving up very few walks and home runs. That fact and the quality of pitchers that appear here, throws a major wrench in the assumption that about 50% would end up with a BABIP at or than the average major league pitcher would have. In my opinion, these pitchers were too talented to assume that they would have high BABIP's put up against them.
Everyone in this sample finished that given season with a FIP- below 100, and three seasons posted by David Wells are the only ones which resulted in a FIP- at or above 85. 40 of the 59 pitchers finished in top-5 in FIP in seasons they qualified for this list (67.8%) and 52 of 59 finished in the top-10 (88.1%). Maybe this is to expected, as HR's, BB's, and K's are the main inputs into the FIP equation; thus you would expect pitchers with a good K/BB and low HR's to have an above-average FIP. But take the time now to scroll through the list of pitchers and just look at the names of these guys. They're the best pitchers of our generation.
So I took another statistic, Fangraphs' WAR, and looked at the quality of these seasons based on that metric. 42 of the 59 pitchers (71.2%) finished in the top-5 in fWAR, a higher percentage than the top-5 based on FIP, while 50 of 59 finished in the top-10 (84.7%). The sample, as stated above, was based on 15 seasons of baseball; 13 of the 15 pitchers with the best WAR in that given season appear on this list. Yet, seven of those 13 WAR (53.8%) leaders posted BABIP's higher than average in that season. However, I realize that Fangraphs uses FIP as their starting point for WAR; thus, again like FIP their WAR may not be the best metric to measure the quality of the seasons considered in this sample, based on the criteria for qualification.
So instead I considered Baseball-Reference's WAR because their starting point is runs allowed as opposed to FIP. The number of pitchers who finished in the top-5 in bWAR dropped fairly significantly from 42 to 29 (71.2% to 50.9%), but the majority of the pitchers just moved back into the top-10 (50 to 43, 84.7% to 72.9%). The quality of pitchers in this sample passes the eye test and looks solid based on a very good metric that doesn't have a DIPS as its starting point.
This sample becomes that much more intriguing when the traditional pitching statistic, ERA, is taken into consideration. In terms of FIP, this is an elite sample of pitching seasons, but in terms of ERA? Not so much. Only 27 of the 59 finished in the top-5 for ERA (45.7%) and that number only increased by five seasons, 32 of 59 (54.2%) when looking at the top-10 on leaderboards. The majority of these pitchers' seasons were still top-10, in terms of ERA, but only by a slim margin, and that majority pales in comparison to the pitchers who were in the top-10, in terms of FIP (a 30.5% difference in fact). My best guess for this over 30% gap, is higher than average BABIP's.
But to get back to my original premise, you wouldn't expect pitchers who were this good in terms of two different measures of WAR to be just average when it comes to BABIP, and I especially don't think you'd expect the majority of them to be below average at it. So are these "bad" BABIP's due to more hittable (but not HR hittable) balls appearing in the zone to hitters, like the original hypothesis predicted? I'm not really sure that a sample size of only 59 seasons is large enough to give conclusive evidence either way.
One thing I will say though, just to further establish the validity of this sample, is that only four of the 30 with BABIP's that were at or above league average in the sample had a lower BABIP than their team's BABIP, during that season. Thus, I think it'd incorrect to say that these pitchers were very good, but bad defenses behind them caused their BABIP's to rise above the league average.
I would just like to conclude with a couple of other slightly interesting findings that came from this study. Schilling was the poster boy who caused Tango to write the blog post that inspired this study, and he did not disappoint. Schilling appeared on the list the most times (5) and also had the most seasons with a higher than average BABIP (4).
Greinke, the reason for my original post, only appeared one time, posting a higher than league average BABIP in that season, but as I stated above Greinke has many more chances to produce Schilling-esque seasons in his career. If Schilling was this study's poster boy, than Greg Maddux was his antagonist. Maddux appeared on the list four times, and in all four of those seasons Maddux had a BABIP lower than league average.
What makes the Schilling and Maddux comparison that much more intriguing, is their respective strikeout and home run rates during the seasons they qualified for this list. Schilling's combined K-rate for those five seasons was 10.18 and his HR-rate was 0.89; while Maddux's K-rate, based on four seasons, was 6.93 and his HR-rate was 0.59. Schilling had a higher strikeout rate and more balls left the ballpark against him; thus, you would assume his BABIP would be lower than Maddux's. But, of course, it wasn't. Maddux's combined BABIP, over his four seasons, was .275; while Schilling's was .298. Maybe, Schilling was worse at suppressing balls in play, or maybe Maddux was more adept at inducing weak contact, it's really tough to tell.
The final interesting finding from the study, came from another great pitcher who has since retired, Pedro Martinez. Many who study baseball statistics assume that BABIP can fluctuate year to year, based on luck (or unexplained randomness), defense, and changes in talent level. Maddux, and Johan Santana as well, were examples of pitchers on this list who consistently were able to have BABIP's lower than average, while Schlling was on the opposite side of the spectrum. Martinez settled nicely in the middle, and showed fairly well that BABIP's still have a massive randomness factor.
Take for instance Martinez' 1999 and 2000 campaigns. In '99 Martinez posted the fourth highest BABIP in the sample (.323), the next season; however, he posted the lowest BABIP in the sample (.236). That is a serious decrease in average on balls in play, but he saw no huge change in his peripherals. There's a chance more balls were put in play against Pedro in '99, as his HR-rate was significantly lower 0.38 to 0.71,but the difference in his K-rate (13.20 in '99 to 11.78 in '00) could have offset that change.
These differences should not lead to an almost 90 point drop in BABIP. There was also no real change in team defense between those two seasons for the Red Sox as their staff BABIP in '99 was .287 and in '00 it was .288. The only plausible explanation for this crazy fluctuation in BABIP, that I can come up with? Randomness, of course.
All Statistics came from Baseball-Reference and Fangraphs.