Limiting Hard Contact: NL Leaders, and a Laggard

Much of modern sabermetric thought regarding pitcher evaluation has been based upon the theory that most types of contact are created somewhat equally. High and low BABIPs allowed are usually attributed to good and bad luck, and FIP, which is directly based upon BABIP, is oft cited as the go-to individual pitching statistic. Well, not all contact is created equal. This week, we’re using a fairly basic method of evaluating contact management ability, and looking at the leading contact managers in both leagues. As it turns out, there’s a head-to-head battle for supremacy in both the AL and NL.

This upcoming weekend, I will be giving a presentation about the best contact managers of all time at the Saber Seminar in Boston. These hurlers were identified utilizing a fairly simple method. Simply strip away all of the strikeouts and walks from every ERA-qualifying starting pitcher’s record. Take the remaining results allowed, assign run values to all of them, and scale each pitcher’s performance on all balls in play to the league average. The resulting figure is the pitcher’s unadjusted contact score. In any given year, there may be a great deal of noise in an individual pitcher’s unadjusted contact score – team defense, ballpark, luck, etc., they’re all in there. Over a pitcher’s career, however, the good contact managers manage contact well, and the bad ones, well….don’t. I’ll leave the history for this weekend in Boston, however – for now, let’s focus on 2014. Yesterday, we looked at the American League – today, it’s the National League’s turn.

In yesterday’s AL article, a table listing the unadjusted contact score leaders in both leagues from 2000-13 were listed. It included names ranging from inner circle all-time greats such as Roger Clemens and Pedro Martinez to relative non-entities such as Joe Mays and Odalis Perez. In between those poles, however, were a significant number of solidly above average major leagues hurlers who relied primarily on contact management rather than on maximization of K’s and minimization of BB’s for their success.

Popup inducers such as Jered Weaver, Jeremy Hellickson, Chris Young and early-career Barry Zito have been league leaders, as well as ground ball generators such as Tim Hudson, Chien-Ming Wang, Russ Ortiz and Derek Lowe. The 2000-13 league leaders’ unadjusted contact scores have ranged from a low of 56 (Derek Lowe – 2002 AL) to a high of 79 (Justin Masterson – 2013 AL). The 2014 AL frontrunners to date, profiled yesterday, are Garrett Richards (59) and 2009 AL leader Felix Hernandez (65). Today, we’re going to take a look at the two 2014 NL frontrunners, through August 8, as well as a laggard who threatens to put up one of the worst single-season unadjusted contact scores in recent memory.

The leaders:
Johnny Cueto – .259 AVG-.396 SLG, 63 Unadjusted Contact Score
Adam Wainwright – .272 AVG-.382 SLG, 65 Unadjusted Contact Score

And the laggard:
Edwin Jackson – .374 AVG-.607 SLG, 140 Unadjusted Contact Score

Historically, both Cueto and Wainwright’s marks are very worthy league-leading scores. Cueto’s 63 matches Jose Fernandez‘ 2013 NL-leading mark, with Wainwright just a short distance behind. We’ll talk a little more about Mr. Jackson later.

As do most top pitching prospects, Cueto diced through the minor leagues on the strength of his ability to maximize strikeouts and minimize walks – managing contact was an unnecessary afterthought as he posted a 378/88 K/BB ratio in 370 2/3 minor league innings, but only a 3.35 ERA over that span. His unadjusted contact scores in his first two ERA-qualifying seasons in the major leagues were quite poor, 120 and 110 in 2008 and 2009, respectively. He improved to 98 in 2010 and 88 in 2012, and is poised to improve by at least 10 basis points for the fifth consecutive qualifying season at the beginning of a career, which has to be some sort of record. Though his fastball velocity is basically the same now as it was at the beginning of his career, his ability to locate it has dramatically improved, and along with the emergence of his changeup, has driven his growth as a pitcher.

Wainwright’s minor league track record is actually quite similar to Cueto’s. He posted a 781/244 K/BB ratio in 793 minor league innings, but his 3.78 ERA didn’t match up to his peripherals by an even greater magnitude than his Reds’ counterpart. Wainwright apprenticed for a full season in the Cards’ bullpen, and qualified for his first ERA title as a starter in 2007. He has managed contact fairly well since, posting a career 90.9 unadjusted contact score entering this season, peaking at 79 in 2010, and bottoming out at 104 in 2012. Wainwright’s ability to generate grounders has been at the core of his contact management success, but as we shall see, he has moved away from this season. He has ridden two legitimate out pitches – first his curve, and later his cutter to his place among the game’s elite starting pitchers.

Now that we’ve identified these two as the leading 2014 NL contact managers based on the raw, unadjusted numbers, let’s hold them up against the scrutiny of context. Are these two guys legitimately elite contact managers, or have they had some help along the way? Let’s review their 2014 plate appearance outcome frequency and production by BIP type data for some clues. First, the frequency information:

FREQ – 2014
Cueto % REL PCT
K 25.8% 127 84
BB 6.7% 87 51
POP 7.2% 94 55
FLY 28.5% 102 69
LD 16.7% 80 1
GB 47.6% 109 79
———— ———— ———– ———–
Wainwright % REL PCT
K 20.6% 101 41
BB 6.0% 78 35
POP 6.0% 77 24
FLY 29.6% 106 80
LD 24.5% 118 99
GB 40.0% 92 21

The most noticeable item on Cueto’s frequency profile is the incredibly low liner rate he’s allowed this season, good for a percentile rank of 1. Early in his major league career, Cueto was a fly ball pitcher, and not a particularly good contact manager overall. More recently, he has been more of a ground ball pitcher, and a much better contact manager. In 2014, his liner rate is so low that he has above average percentile ranks in all of the other BIP categories. He still has a bit of a grounder tendency (79 percentile rank), but it not as extreme as in recent seasons. His low liner rate should regress moving forward – in his five previous qualifying seasons, his liner rate was below league average twice, above twice, and right on the average once. His K rate has taken a strong step forward this season (85 percentile rank), without material deterioration in his BB rate (51). Missing more bats plus an off-of-the-charts low liner rate equals career year.

Wainwright’s frequency profile is, for lack of a better term, pretty weird. If you would have shown it to me before the season and said that it would be Wainwright’s in early August, I would have assumed that 2014 would have been the year of his great decline. Plunge in K rate (to percentile rank of 41)? Check. Upward bounce from career-low 2013 BB rate to previous level (35 percentile rank)? Check. Evaporation of his longstanding significant ground ball tendency (to 21 percentile rank)? Check. Explosion of liner rate to a higher mark than anyone posted in the NL in 2013 (99 percentile rank)? Check. Somehow, some way, Adam Wainwright has had a downright Wainwright-esque year despite all of this.

We’ll learn more below, as we take a look at the production by BIP type allowed by both pitchers, both before and after adjustment for context, to get a better feel for the batted-ball authority they have allowed:

PROD – 2014
Cueto AVG OBP SLG REL PRD ADJ PRD ACT ERA CALC ERA TRU ERA
FLY 0.229 0.568 63 77
LD 0.623 0.783 85 99
GB 0.203 0.223 71 71
ALL BIP 0.259 0.396 63 72
ALL PA 0.185 0.242 0.283 58 64 2.05 2.13 2.36
————- ———- ———- ———- ———- ———- ———– ———– ———–
Wainwright AVG OBP SLG REL PRD ADJ PRD ACT ERA CALC ERA TRU ERA
FLY 0.201 0.403 38 46
LD 0.622 0.793 86 88
GB 0.155 0.182 44 85
ALL BIP 0.272 0.382 65 80
ALL PA 0.211 0.259 0.296 65 79 2.34 2.39 2.88

The actual production allowed by both pitchers on each BIP type is indicated in the AVG and SLG columns, and is converted to run values and compared to MLB average in the REL PRD column. That figure is then adjusted for context, such as home park, team defense, luck, etc., in the ADJ PRD column. In the three right-most columns, their actual ERAs, calculated component ERAs based on actual production allowed, and “tru” ERAs, which are adjusted for context, are all presented. For the purposes of this exercise, SH and SF are included as outs and HBP are excluded from the OBP calculation.

The core of Cueto and Wainwright’s contact management success is rooted in their ability to limit fly ball damage. Cueto has posted a 63 unadjusted contact score on fly balls, Wainwright an outlandish 38. There are two ways to successfully limit damage on fly balls – to manage the vertical angle and the exit velocity off of the bat. Splitting the fly ball category into upper and lower groups equidistant in size from the popup and line drive borders yields starkly different results. The “higher” fly balls yield an .098 AVG-.234 SLG, while the “lower” ones yield a .380 AVG-.990 SLG. Roughly 35% of fly balls reside in the upper group, 65% in the lower.

Cueto has limited damage on fly balls by allowing significantly more “high” fly balls than the average pitcher. 47.5% of the fly balls allowed by Cueto to date have been “high” fly balls. He has allowed higher than MLB production (.089 AVG-.339 SLG) on them, as Great American Ballpark is often quite windy and high fly ball-friendly, while allowing a .355 AVG-.774 SLG on “low” fly balls – below league average, but not to an extreme, elite extent. He has induced these high flies with his entire array of pitches, but his changeup and cutter have been the two most successful offerings. Cueto has suffocated fly ball production by managing the exit angle more so than by limiting authority, and one should expect his “high” and “low” fly rates to normalize in future seasons, especially since he doesn’t have a track record as a popup guy. He’s living on a razor’s edge with regard to fly ball contact this season – his contact score is adjusted upward to a still strong 77 based on context – and likely won’t remain there.

Wainwright, on the other hand, has a quite low “high” fly ball rate, 29.9% thus far in 2014. He strangles fly ball production in both sectors, however, allowing .100 AVG-.200 SLG in the upper tier, and .245 AVG-.489 SLG in the lower. That paltry level of “low” fly ball production is quite amazing. His fly ball contact score is barely adjusted upward for context to 46, as his hard/soft fly ball rates are exceptional. Wainwright’s curve and cutter have been his foremost weapons in generating weak fly ball contact.

Both pitchers have allowed weaker than MLB average authority across all batted ball types, but it should be noted that all of the contextual adjustments for both pitchers increase their component contact scores. Most significantly, Cueto’s line drive contact score increases from 85 to 99, and Wainwright’s grounder contact score spikes sharply from 44 to 85, as his .155 AVG-.182 SLG yielded on grounders has a healthy dose of luck in it. Despite all of this, both pitchers have better than MLB average adjusted contact scores on all batted types, though their overall contact scores on all BIP increase to 72 (for Cueto) and 80 (for Wainwright) once adjusted for context. After adding back their K’s and BB’s, their “tru” ERAs are 2.36 for Cueto, and 2.88 for Wainwright, 0.39 and 0.34 better than their actual ERAs. Both are exceptional pitchers having exceptional years, but are pitching a bit above their numbers.

There is one guy who could still catch them in the race for NL Contact Manager of the Year. That would be Clayton Kershaw, whose unadjusted contact score has plunged from 105 to 86 in the last five weeks. His adjusted contact score in early July was 84, and is likely better than either Cueto or Wainwright’s as we speak. Kershaw is an elite all-around pitcher – a top-of-the-scale bat-miser with unmatched contact management skills – his 73.9 career unadjusted contact score is an all-time record for a pitcher’s first five qualifying seasons. Cueto is a good, improving pitcher having his career year, Wainwright is showing unmistakable signs of decline, but is busily compensating in other areas, as only as the really good ones can.

What about Edwin Jackson? Here are his frequency and production tables:

FREQ – 2014
E.Jackson % REL PCT
K 20.1% 99 38
BB 9.6% 125 95
POP 6.8% 88 46
FLY 28.1% 101 64
LD 26.5% 127 99
GB 38.6% 89 17
PROD – 2014
E.Jackson AVG OBP SLG REL PRD ADJ PRD ACT ERA CALC ERA TRU ERA
FLY 0.404 0.981 191 123
LD 0.684 0.898 106 107
GB 0.203 0.238 75 100
ALL BIP 0.374 0.607 140 128
ALL PA 0.288 0.357 0.468 139 128 5.61 5.10 4.70

Jackson has always thrown hard, and has been coveted – and paid well – by club after club. While durable, the flaws highlighted in the above tables have always been present. His liner rate has been above MLB average in six of the last seven seasons, and is in the stratosphere this season. Not only has his longstanding grounder tendency dried up this season, he has allowed well above average authority on fly balls and liners this season. Allowing a .404 AVG-.981 SLG on fly balls – a 191 unadjusted fly ball contact score – is beyond the limits of good taste. Adjustment for context brings it down quite a bit to 123, but that’s still way too high for a guy who allows a lot of fly balls. The only positive in his production profile – his 75 unadjusted grounder contact score – is watered down by adjustment for context, as well as the fact that he hasn’t yielded many ground balls this season. His 5.61 actual ERA is boosted a bit by both sequencing and context, but his “tru” ERA of 4.70 places him right around replacement level, where he belongs.

Jackson’s 140 unadjusted contact score is in pretty select company. Only four pitchers – Brandon Backe (149, 2008), James Shields (149, 2010), Ivan Nova (146, 2012) and Jose Lima (140, 2000) have posted unadjusted contact scores of 140 or worse this century. All of this – both positive and negative – is simply more evidence that not all batted balls are created equal. There are different ways of getting it done, either by inducing larger than normal amounts of popups, grounders, or even “high” fly balls, or “low” ground balls – or just by minimizing batted-ball authority across batted-ball types. The good contact managers find a way to get it done, while the bad ones don’t.




Print This Post





21 Responses to “Limiting Hard Contact: NL Leaders, and a Laggard”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Grumpy Ole Bass Turd says:

    “Much of modern sabermetric thought regarding pitcher evaluation has been based upon the theory that most types of contact are created somewhat equally. ”

    That’s pretty much exactly NOT modern sabermetric thought. Or ancient sabermetric thought. Or any variety of sabermetric thought except “misconstrued”. The principle you’re mangling is that the park-defense-opponent adjusted variation in the results of GBs allowed and non-HR airballs allowed is extremely random and not very predictive over a year (or the 75% of a year here).

    Vote -1 Vote +1

    • Wobatus says:

      I don’t think he’s mangling it. He’s right (speaking in shorthand) that FIP and xFIP (subbing in for sabermetrics) consistently seems to overate Jackson and underrate guys like Chris Young, Weaver etc. FIP and xFIP don’t even take infield flyballs into account. He’s saying pretty clearly that the results of GB allowed and non-HR airballs allowed is less random than you assume.

      Vote -1 Vote +1

      • Grumpy Ole Bass Turd says:

        Of course he’s mangling it. Nobody thinks a smashed grounder and a weak grounder are the same or have the same run expectancy. Nobody. And FIP for fWAR has included IFFB as equivalent to K for awhile.

        Vote -1 Vote +1

        • Wobatus says:

          It doesn’t say so in the glossary re FIP so I was unaware. And you can’t say that nobody thinks x and you yourself say the results of ground balls are random. which suggests that the number of smashed as opposed to weak is random if you also believe the results for both indeed aren’t the same.

          Vote -1 Vote +1

        • Grumpy Ole Bass Turd says:

          I never said the results of ground balls were entirely random. Strawman much? And even supreme absolute god of contact management Clayton Edward Kershaw allows a BA of .215 on grounders compared to a .233 league average over the same timeframe, and he has pitched in front of well above average IF defenders. Allowing easily-converted grounders is just not *much* of a skill.

          And when the article talks (obliquely) about Cueto and his .180 BAA on grounders this year (.220 career), without mentioning that his difference from league average this year is something like 90% noise+defense and 10% skill, it’s somewhere between disingenuous and ignorantly misleading.

          Vote -1 Vote +1

        • Wobatus says:

          You said exactly that. The results of ground balls allowed is extremely random. Direct quotation. I think stronger ground is fly balls. Hr/fb includes iffb.

          Vote -1 Vote +1

        • Wobatus says:

          Ok, you said extremely random and not entirely. Hardly a straw man though.

          Vote -1 Vote +1

        • Grumpy Ole Bass Turd says:

          Exactly. And he’s quoting stuff numbers that are maybe 10% skill as though they’re.. I dunno, meaningful in any real way. Only a tiny difference between Cueto/Kershaw and an average pitcher on the same team comes suppressing 1 hit per every 100 GB allowed, or if you’re EJax, giving up 1 extra hit per every 100 GB allowed. Pretty sure you’ll give me that one, and I’m not convinced that the true skill is even +/- 1 hit per 100 GBs.

          I agree that limiting the FB damage profile via IF/(IF+OF) ratio is a very real pitcher skill. But once you subtract out the IFFBs and treat them like Ks, how much predictive information is left in 3/4 season of the (park/defense-adjusted) FB damage numbers? Not a ton. Once you pull HRs out too, how much predictive information is left in 3/4 season of the (park/defense-adjusted) FB damage numbers? Very little.

          Vote -1 Vote +1

        • Wobatus says:

          Yeah, GOBT, if you treat the pop-ups like Ks (I didn’t realize that), and/or took them out of FBs altogether, er left with just putting a very fine coat of gild on the lily. Good points. Uncle.

          Vote -1 Vote +1

  2. Robert Hombre says:

    Same comment as yesterday: you’ve done an excellent job of describing what has occurred. Has. Past Tense. If you’re only striving to describe what has occurred, excellently and meticulously done, and read no further.

    If, however, these articles are supposed to indicate that anything about RELPRD on balls in play is predictive, then you can’t just fiat that. That’s how we get ESPN’s esoteric, proprietary numbers (their football QBR, for instance) that are caged behind a steel box of mystery, numbers that no one really gives credence to because they aren’t open to critical examination and dissection.

    I’m not implying that you’re wrong or irresponsible in not disclosing RELPRD stabilization rates. We know that ground balls stabilize in about 100 BIP. We know that Line Drives stabilize around 2000 BIP. That’s the point at which a pitcher’s own GB% becomes more predictive than the mean. We know this because these numbers aren’t hidden behind ESPN’s proprietary Google-PRISM-Obamacare Wall of Secrecy. We know that GB% is predictive after 100 BIP because someone did the work to determine stabilization rates. When do RELPRD rates stabilize? 100 BIP? 1000? 5000?

    Until the writhing masses see stabilization rates, these articles are extremely interesting dissections of what has occurred. Nothing less than that, but nothing more.

    Vote -1 Vote +1

    • Wobatus says:

      Edwin Jackson has mostly been worse than his xFIP etc. Yes, it varies some. Young, Weaver etc have been better. Over a longtime. Yes, it varies as well. xFIP doesn’t take IFFB% into account at all, although it’s a clear skill. It also uses K per IP and not K%. Using those may gild the lily some, but clearly there is some aspect of contact quality that doesn’t get into the standard saber metric measures, things that don’t seem all that random.

      I think Tony is adding to the toolbox and folks seem a tad offended, like he is an apostate.

      Vote -1 Vote +1

      • Robert Hombre says:

        When I said that he described the past excellently, that wasn’t me being sarcastic. This is a contribution, a substantial one.

        But he’s used Relative Production on batted ball types as predictive measures before, so stabilization numbers (which, as novaether noted, I explained with all the nuance of a B-52) seem paramount. It’s just surprising to me that no one’s asking that Tony show his work IF he’s pushing this as a predictive statistic.

        Like I said, if Tony’s not implying that, if he’s saying that it’s just a retrospective, then it’s a fine contribution to the baseball internet. But if it’s being implied that relative production on batted ball types is a pitcher skill, I want to know when it becomes reliable, so that I know how it’s used responsibly.

        What’s unreasonable about that?

        Vote -1 Vote +1

        • Wobatus says:

          Yeah, mine was a poor response to your post.

          That said, to me the data is clear we’ve relied too much on the ideas for example, that home run rates in fly balls are random.

          Take Chris Young. hR/FB uses all fly balls and doesn’t exclude iffb. So he consistently has low hr/fb rates. He also has extreme low gb rates. And low ld rates. His career line drive rate would make him 15th lowest of all qualified starters this year. He clearly is not a random chance outlier. He has skills that enable good results xFIP and FIP can’t recognize. All fly balls aren’t created equal. To a large extent, yes, and it would be great to see stabilization data, but Young is at such an extreme, Weaver to a degree, that it pretty much stares you in the face. Same with Tom Glavine etc.

          Vote -1 Vote +1

        • Robert Hombre says:

          And so to continue along that line of reasoning, Chris Young’s batted ball profile is excellent and it’s entirely reasonable to believe his fly-ball mastery, IF/FB as substantially greater than HR/FB, is a skill. It’s what separates him from equally-flyball prone but notably less successful Josh Tomlin (whose xFIP is 3.14, to the bewilderment of all).

          This is the case for Weaver and Young because they’ve been pitching in the league for a decade, and yes, their BIP Relative Production is apparent because the trends have been there so long. They do stare you in the face. They are the extremes.

          If you would like to change your examples, I won’t hold it against you or try to trap you into a corner. However, what the use of Weaver and Young as examples would suggest – and if you’re not saying this, please correct me – that it’d only be after nearly a decade that one could conclude. It’d have to be a case-by-case basis of pitchers over 30, and by that point, BIP relative production takes on a very narrow predictive latitude, perhaps ten to twenty pitchers whose peripherals and defense-normalized RA9 don’t match up.

          That’s why stabilization numbers aren’t just trivia. It’s not to discredit Tony’s assertions; to the contrary, in fact. Stabilization numbers would help us expand his analysis beyond just the fringe cases like Weaver and, depending on how small the stabilization rate, would allow us to look at perhaps smaller-sample pitchers, the Doug Fisters or Corey Klubers, and allow us to more fruitfully and critically analyze their BIP profiles to distinguish signal from noise. If it’s quick to stabilize like K% or GB%, we could identify the laggards more quickly; if it turns out the stabilization numbers are massive like LD% (2000 BIP to stabilize!), then it could teach us to exercise caution when analyzing BIP relative production.

          Fair?

          Vote -1 Vote +1

        • Wobatus says:

          Yup, there’s a fair cop.

          Vote -1 Vote +1

    • novaether says:

      Stabilization =/= the point at which a player’s own rate becomes more predictive than the mean. It’s the point at which more than 50% of the variance of its outcomes can be explained by something other than random variation. It’s a retrospective tool commonly misused by nearly everybody. The reason it’s not prospective is that the “something other than random variation” (player’s skill, the ballparks he played in, the opponents’ skill, etc) have a tendency to change.

      Vote -1 Vote +1

  3. Bat-miser says:

    I didn’t realize that Clayton Kershaw was buying all the bats and hoarding them in his garage.

    Vote -1 Vote +1

  4. novaether says:

    Where are you getting your data? I highly recommend you disclose this in every article you write, unless it’s obvious, for future reference.

    Vote -1 Vote +1

    • The Captain says:

      Judging from the crazy detailed data he has access to and the fact that, despite writing only terrific pieces of analysis, he has never cited his sources, I always assumed Tony was under some sort of NDA.

      Vote -1 Vote +1

      • novaether says:

        Good sabermetrics is science. Good science cites its sources. It’s a dramatic example, but would you want your doctor writing you prescriptions to medications that claim effectiveness without releasing a formal study?

        Vote -1 Vote +1

        • a eskpert says:

          Good sabermetrics can also be journalism, depending on how ephemeral it is, and journalism often requires anonymous sources.

          Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>