Challenge: Prove Brandon McCarthy’s HR/FB is Not Bad Luck
In two starts spanning 12 innings, Brandon McCarthy has allowed six home runs. Out of 12 fly balls. For a 50% HR/FB rate. After witnessing the last of those homers in Monday’s game, I Tweeted about it. Because that’s what all the cool kids do nowadays, Tweet, right? After a couple of back and forths, I then Tweeted the following:
@RotoLance Anyone who claims to have evidence for why McCarthy, or any pitcher, has allowed an inflated HR/FB rate is lying or just wrong.
— Mike Podhorzer (@MikePodhorzer) April 14, 2015
I stand by that statement. We’ve come a long way in the past decade in understanding pitcher performance. With the introduction of PITCHf/x, we know far more than we have ever known before. But there still remains a wealth of knowledge that I wish we had.
When Voros McCracken first introduced the world to DIPS, most understandably scoffed. Eventually us statheads embraced his ideas and metrics such as FIP that were built upon those ideas were created. Then when we discovered that a pitcher’s HR/FB rate was quite flaky from year to year, the xFIP metric was developed. With all these metrics drilling into our heads that this is what a pitcher’s ERA should be absent luck on balls in play and sequencing randomness, we perhaps went too far, believing that any difference between a pitcher’s ERA and DIPS ERA is due to good/bad luck. Now, it has become trendy to acknowledge the luck component, but then side-step it and try justifying the out/underperformance.
The problem with this is that there is never any evidence provided to support what the scout in you supposedly sees. You say that Pitcher X throws too many meatballs? Okay, prove it. Cite facts, data, evidence. Compare X’s meatball percentage to every other pitcher. Is it actually higher? Do his meatballs get launched for homers more frequently than other meatball throwers? If so, why?
Without this research, you’re just succumbing to confirmation bias. Rather than diving into the problem without any preconceived notions, the analyst points to an inflated (or suppressed) HR/FB rate as the evidence that a pitcher is “too hittable”, “throws too many strikes/around the plate too often” (my favorite) or whatever description is decided upon. An inflated HR/FB rate on its own cannot be the evidence. How do we know that it’s not bad luck?
McCarthy’s unfortunate start is simply the continuation of his struggles with the long ball that began last season. His 16.3% HR/FB rate easily led all qualified Major League starters. But consider this — before last year, McCarthy’s career HR/FB rate sat at 9.3%. That actually rates below the league average.
So if you lean toward McCarthy lacking the skill to prevent home runs on fly balls, then what has changed since 2013? It’s true that McCarthy is actually a very different pitcher now. But his transformation should have boosted his ability to prevent fly balls from leaving the park, not reduced it. He has morphed from soft-tossing flyballer into a hard-thrower with a ground ball tilt. And fastball velocity is negatively correlated with HR/FB rate. So any argument that McCarthy’s new found velocity is increasing the opposing batter’s speed off the bat, resulting in more homers, is a flawed one. Maybe you think his hotter heat is straighter than it used to be and straighter fastballs lead to more homers? Okay, fine. Interesting hypothesis. Go test it!
Now let’s get this out of the way. I am by no means suggesting that I naively believe any pitcher whose HR/FB rate deviates from the league average is lucky/unlucky. I’m merely conceding that although luck could be the primary culprit, perhaps it’s a difference in skill. I just don’t know for sure. No one knows.
Clayton Kershaw owns a career HR/FB rate of just 6.7%. We all know he is the best pitcher in baseball and has been a member of the elite for many years now. But what about Chad Billingsley? In 1,175.1 innings, he has posted a suppressed 7.6% mark! Both have called Dodger Stadium home their entire careers, but sport nearly identical home/road splits, so it’s not a ballpark thing. Clearly in over 1,000 innings, the possibility that luck plays a major role is severely reduced. So it’s obvious that some pitchers do seemingly have the skill to limit home runs on fly balls, while others don’t. But how do we measure, quantify and understand it? How do we know?
So I am issuing a challenge:
Prove that Brandon McCarthy’s HR/FB Rate is Not Just Bad Luck
I am admittedly devoid of the necessary skills to research this on my own. I wouldn’t even know where to begin. So dive on in and share your findings. Any claims you make should be supported by facts. Even if you come to the conclusion that it is all bad luck, I want to know!
For those that accept my challenge and perform the research, send me an email (in my signature). I will publish a follow-up and share anything interesting so we could continue this discussion. It’s your chance to see your name in lights in a FanGraphs article!
Mike Podhorzer is the founder of ProjectingX IQ, an advanced fantasy baseball analytics platform that transforms projection data and in-season performance signals into actionable intelligence. He is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year and three-time Tout Wars champion. He is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. Follow Mike on X@MikePodhorzer and contact him via email.
First hypothesis: GB / FB Splits (see yesterday’s Daily Grind)
Fly ball hitters account for the majority of home runs, and they hit better against ground ball pitchers. Like Mike, I don’t have the database skills to verify this. I’m pretty sure it’s the answer, although there might be a slice of bad luck baked in too.
To clarify, I should say I’m pretty sure it’s part of the answer. There are probably competing effects. McCarthy is definitely reaching the point where his elevated HR/FB looks like a skill rather than noise.
I’m confused. Your post just talked about hitters vs pitchers and hitter splits vs GB/FB pitchers so how does this relate to McCarthy’s HR/FB rate? Are you suggesting that he has faced an inordinate number of fly balls hitters who have accounted for the majority of the homers since the beginning of last season?
I’m suggesting his profile is inordinately good for fly ball hitters. He doesn’t have to face an above average rate of fly ball hitters for this to be part of the answer/problem.
http://www.fangraphs.com/leaders.aspx?pos=all&stats=pit&lg=all&qual=y&type=8&season=2015&month=0&season1=2013&ind=0&team=0&rost=0&age=0&filter=&players=0&sort=15,d
Top 7 guys on the list, 6 of them are extreme ground ball guys (Danks is just a terrible pitcher)
http://baseballsavant.com/pitchfx_search.php?hfPT=ff%7Cft%7Csi%7C&hfZ=&hfGT=R%7C&hfPR=13%7C&hfAB=17%7C&ddlStadium=&hfBB=&hfHL=&pid%5B%5D=433584&pid%5B%5D=572971&pid%5B%5D=435221&pid%5B%5D=489119&pid%5B%5D=503449&hfCount=&ddlYear=&ddlPlayer=pitcher&ddlMin=0&ddlPitcherHand=&ddlBatterHand=&ddlVGT=&ddlVLT=&ddlBBVGT=&ddlBBVLT=&ddlDistGT=&ddlDistLT=&txtAngleGT=&txtAngleLT=&txtGameDateGT=2013-01-01&txtGameDateLT=&ddlPlayerHeightGT=&ddlPlayerHeightLT=&ddlPlayerWeightGT=&ddlPlayerWeightLT=&ddlTeam=&ddlPosition=&hfRO=&ddlHomeRoad=&hfIN=&hfOT=&ddlGroupBy=name&ddlSort=desc&ddlMinABs=0&ddlSBSuccess=&txtPx1=&txtPx2=&txtPz1=&txtPz2=&ddlRPXGT_ft=&ddlRPXGT_in=&ddlRPXLT_ft=&ddlRPXLT_in=&ddlRPYGT_ft=&ddlRPYGT_in=&ddlRPYLT_ft=&ddlRPYLT_in=&txtBAGT=&txtBALT=&txtBLGT=&txtBLLT=&txtSRGT=&txtSRLT=&txtSDGT=&txtSDLT=#results
a lot of fastballs up in the zone from these 5 (see “pitch graph”, the 2nd to last column on the right for each pitcher)
also in regards to mccarthy
2012 32% sinkers 8.1% HR/FB
2013 43% sinkers 10.3% HR/FB
2014 51% sinkers 16.3% HR/FB
There is selection bias going on here; if you’re a FB pitcher and have a high HR/FB rate, you’re not going to be in the league very long.
Nick That is survivorship bias, and you are certainly right
I agree with this as much as I can, given the question. Although, I think it’s probably a result of the second part of your answer, or the evidence, rather. From this point on, just preface everything I say with, “I think….”
Basically, it’s not the batted ball tendency, it’s the pitches thrown to get the batted ball and their placement.
When you look at the guys that generally perform poorly in this regard over the long haul, you’re looking at guys with a fastball/two-seam/sink repertoire. I would imagine that this also has a lot to do with why velocity correlates to HR/FB — A lot of these guy’s FA’s are really just FT or SI that are borderline.
Maybe if you query “borderline” FT/SI’s you’ll find a correlation. The problem with that is, if you just query FA, there’s so many of them, you won’t find anything. And if you query FT/SI all together, you’re probably not going to find much of anything (although i bet you find at least something) — so if you query fastballs with like, i dunno, 4-8 vertical, and maybe 7-10 horizontal?
Honestly, I’m sure the best way to do this is to just take the guys with sustained high HR/FB rates, and plot their fastballs (that resulted in a flyball if you want) on the standard pfx movement graph, and highlight dingers.
My guess is that there’d be at least some grouping in that inbetween zone.
I knew this brought to mind a recent THT post, and here it is:
http://www.hardballtimes.com/batted-ball-trajectory-splitting-the-difference/
Shane didn’t look at HR rates there but he found that while FB hitters do produce higher liner rates against GB pitchers, they actually produce lower than average FB rates of just around 30~31%. And they produce slightly above average GB rates. So essentially they turn some classified FB into liners, which means better contact, but without further information on these liners it’s impossible to know whether it resulted in more HR or not.
2105 Hard Hit Ball % extremely high (58) combined with higher GB/FB split.
Where is this data from? And of course his hard hit ball % has been high, he has allowed six homers already! The question is whether it’s bad luck or a lack of skill in preventing hard hit balls.
There is “never any evidence”, the result is the evidence, whether caused by the mystical good luck or bad luck. What is evidence if one does not consider the result. And if the result cannot be the evidence, how do we know so and so is a crappy hitter, so what he strikes out 35% of the time how do we know it’s not just bad luck? How do we know someone is a strikeout pitcher or a homerun hitter, we can’t offer that they get a lot of strikeouts or they hit a lot of homers as proof? Isn’t that what you are saying here; “the analyst points to an inflated (or suppressed) HR/FB rate as the evidence that a pitcher is “too hittable”, “throws too many strikes/around the plate too often” (my favorite) or whatever description is decided upon. An inflated HR/FB rate on its own cannot be the evidence. How do we know that it’s not bad luck?”. No statistical analysis stands up without consideration for the results. He’s a great pitcher he throws 99 mph and never throws a ball, he has a great arsenal score and so on, but wait are we leaving out batters hit .400 off him, must be luck. Results are the only thing that matters.
It’s the underlying skills that drive the results. The ability of Giancarlo Stanton to hit his fly balls far, averaging over 300 feet that leads to all those long balls. That’s the evidence. The components of my xHR/FB rate equation is the evidence. Not the HR/FB rate itself.
Strikeout pitchers induce swings and misses. That’s the skill. They also get called and foul strikes, those are also skills. They lead to strikeouts. Throwing lots of strikes is the skill that leads to avoiding the free pass.
Getting swings and misses is the result, throwing 99 mph, throwing deceptive pitches with break, throwing with accuracy, those are the skills. Getting swings and misses and getting called strikes is the process that leads to a strikeout, so you too are merely looking at the results.
And that was my point “it’s the underlying skills that drive the results”, so there is currently a problem with McCarthy’s underlying skills at preventing a bad hr/FB rate, if the notion is, so what, I agree, given the small sample size and his history.
No, you take the evidence collected and perform valid statistical tests on them to show whether or not the results are statistically significant.
It’s not JUST bad luck.
It’s bad luck combined with small sample size.
“Without this research, you’re just succumbing to confirmation bias. Rather than diving into the problem without any preconceived notions, the analyst points to an inflated (or suppressed) HR/FB rate as the evidence that a pitcher is “too hittable”, “throws too many strikes/around the plate too often” (my favorite) or whatever description is decided upon. An inflated HR/FB rate on its own cannot be the evidence. How do we know that it’s not bad luck?”
i had the idea once to comb through all the zone profiles of major league pitchers and see which areas of the plate/pitches in certain areas lead to the highest hr/fb. obviously, pitches off the plate would be low and ones over the center would be the highest, with each hitter having different strengths. the other problem is predictability and how much that can lead to an increase in home runs. while you would eventually come up with something like “fastballs in quadrant XX go for home runs at a 20% rate”, it’s still a specific case. if you throw 85, there’s a good chance its higher than 20%, if you throw 100, its probably lower. then if a pitcher throws a large percentage of a certain pitch that has a high hr/fb in a certain zone, there’s your evidence.
Danny’s thoughts are in line with my immediate gut feeling. I also think that merely looking at HR/FB rates is being way to vague, if you’re trying to identify the problem.
Is there an easy filter in any of the new fancy databases? It would be helpful to know all the information that went into all the home runs hit off a pitcher.
Handedness of the batter
HR/FB rate of the batter
Pitch type of home run
Pitch speed of home run
Pitch location of home run
Count of the at bat of home run
Hr distance
Hr hang time
Hr landing location (LF, CF, RF)
*Fly ball distance avg (this is done for hitters, it would be just as useful for pitchers, especially in comparing with Hr/FB distance)
Hr/FB RATES separated by pitch type would be useful as well.
My guess is that some of these would give us the answers to most pitchers. Again, maybe there is no way to look any of that up easily, but I would think one could with all the new data out there. I’m just not that man, and I’m not claiming to be too busy….just too incompetent
Agree with both above, and to further the cause, look into pitch counts to where pitchers are more likely to throw specific pitches.
You will find evidence in a lot of pitchers with specific repertoire on their weaknesses in counts. Batters won’t accidentally be in a count where a slider is more likely to come and hit a homerun.
From this point, you can see that predictability will showcase which pitches are most likely thrown, and the outcome. McCarthy, behind in the count on RH in 2014, had around 5-6% of his pitches in the bottom-away of the zone. 85% made contact, and the slg% per pitch hovers around .300-400.
This year, he’s placed 8-10% of pitches in that zone, where 90% have made contact and the slugging is around 1.000. It seems to point out that maybe skill is involved (but not overall skill, temporarily) that he isn’t hitting the corner enough where he was getting more outs in the spot last year.
More of his pitches are coming closer to the RH, in counts where they are expecting strikes and/or challenge fastballs. Consdering that a few of these zones have slugging% per pitch of 2.000 and 4.000, that evidence in itself doesn’t correlate to his history, and doesn’t correlate to any pitcher that would last in the majors longer than a couple of starts.
If the zones surroudning the bad hit zones had also increased, there would be worry. But those other zones he hasn’t really hit yet behind in the count (ie: not hitting his spots or over-challenging to close to the plate) or have low/expected contact% and slugging% (ie: if he hits his spots behind in the count he can still dominate the hitter.)
Counts would be the best evidence, since from a Sabr and scout view, game-calling can be dissected easier since McCarthy is more likely to throw a certain pitch or zone behind in the count. And in this case, there is a huge increase of “meatballs” thrown in the outside-bottom strike zone that is getting hit hard.
These zones weren’t a weakness of his last year, but the one change is the increase of pitches there, and the decrease of pitches at the edge of the plate. Homeruns might be an outcome of easily predicting his pitches, culminated with his inability to hit his spots.
Skill and luck. He needs to hit the catchers mitt on the edge more often, possibly not assume the pitcher-friendly park is going to save him from trying to get a fastball by hitters, and a 4.000 slugging% at the bottom zone isn’t sustainable even with the latter two issues discussed.
https://yourlogicalfallacyis.com/burdenofproof
I think you’ve got this one wrong. The null hypothesis here is that McCarthy’s *true* HR/FB is league average. So if you claim that his high HR/FB so far this season is anything but random variation then the burden of proof is on you.
I’m no expert, but it seems odd to me that in a 1 on 1 battle of hitter vs pitcher (which is all it is when only considering homers), that the null hypothesis is that only one of the players has any control over a major aspect of that 1 vs 1
I’ve got a theory on this, which I will try to back up with sloppy math. It’s the Rickey Nolasco/Jeremy Bonderman theory. This subset of pitchers is so great at throwing strikes that, when faced with a hitter’s count, they can throw with regularity throw a strike. For pitchers with lesser control, this outcome more often results in a walk. But these pitchers are so good at throwing strikes and so walk-aversed that they can and do put the ball right in the zone and get hit… hard.
In 2014, in the 89 innings in which McCarthy pitched in hitter’s counts (I defined this as 3-0, 3-1, 3-2, 2-0, and 1-0), he gave up 19 HRs.
In the 272 innings (testament to his great control) in which he pitched in pitcher’s counts (I defined these as 0-1, 0-2, 1-2, and 2-2), counts where he could conceivably throw a ball and be okay, he gave up only 10 hrs.
This should be verified with better count data than the fangraphs site, and if possible, with some sort of batted ball data (LD% or hard hit ball%) to really get to the bottom of it. But it seems that he’s so good at throwing strikes that he throws them when he shouldn’t.
We need context here and a comparison to other pitchers as well. Of course hitters are better in hitter’s counts. Has McCarthy allowed a higher rate of homers in those counts than normal? If so, why?
The too many strikes theory I noted and just sounds ridiculous to me. You’re just cherry picking names of strike throwers who have inflated BABIP marks. How about those strike throwers who don’t suffer from such an affliction?
Could be an interaction of stuff and strike throwing. At some point stuff overwhelms predictability.
Just playing devil’s advocate.
You’re double counting a lot of home runs with those numbers. He also allowed 7 first pitch HR that are not captured by your counts so of his 18 remaining HR, you have accounted for a whopping 29!
Per Baseball Reference, He allowed 7 HR on those counts you list as hitter’s counts, 9 if you add 2-1, and allowed 5 on pitcher’s counts. I’m not sure about the best way to express this as a rate stat. As a percentage of total plate appearances that ended on those counts? As a percentage of all pitches thrown on those counts?
For a few quick comparisons, 2014 NYY teammate Hiroki Kuroda gave up 6 HR in hitter’s counts, 7 including 2-1, and gave up 2 HR on pitcher’s counts.
Best-pitcher-in-the-world Clayton Kershaw gave up…1 HR in hitter’s counts in 2014, still 1 if you add 2-1 counts. In pitcher’s counts he gave up 5 HR.
All had extremely similar IP totals. For the remaining HR, McCarthy and Kuroda both allowed 7 on the first pitch and 4 on 1-1 counts. Kershaw allowed 3 and 0.
It seems to me that two related questions are being conflated: (1) is HR/FB a (lack of) skill, and (2) if it is, can we describe the skill any closer to the pitcher than the ball going over the fence. #1 is akin to BABIP — it is an overgeneralization that BABIP is all luck, but it is still usually true — relatively few pitchers have demonstrated year-to-year BABIP, and even then, some of those are still the beneficiaries of randomness. Similarly, HR/FB doesn’t correlate year-to-year for most pitchers — ie. mostly luck — but it is entirely possible for those for which it does, it is the result of skill. If this is what Rico is saying, he’s not wrong. But proving so statistically is even harder than BABIP, though, because as Dolemite notes, the HR sample sizes for any pitcher who has enough skill to stick the ability stay in the league any length of time are necessarily small.
Getting to #2 seems to be Mike’s goal, but this seems nearly impossible given how hard it is to even answer #1. If we could answer #1, we could seem what similarities pitchers who demonstrate consistency share, and hypothesize from that about the “skill.” But any attempt to do so is likely to be confounded by the fact that, given the sample sizes, some of those included in the sample are beneficiaries/victims of luck.
It seems to me that StatCast could help produce at least a partial answer by increasing the sample size. HR/FB is (mostly) the result of the distance of each FB, which is (partly/largely) the result of launch angle and velocity. If a pitcher shows strong tendency to generate batted balls with angle/velocity/distance in identifiable, problematic “zone(s),” then it’s less likely to be luck that many of those go over the fence. If a pitcher’s batted balls are more randomly distributed, then it’s more likely to be luck. By categorizing every batted ball, we get a much larger sample.
If we are able to identify which pitchers give up problematic batted balls, we might be better able to generalize about why.
Kimball, great summary
Ya more Statcast would be awesome. If pitcher X had a much higher exit velocity on pitches in the zone over a long enough sample size this would strongly argue to be skill(or lack thereof) based vs pure dumb luck
Thanks for the well thought out response. This is precisely where I’m going with this. Basically with our data, we can’t get to the proof. It bothers me then that some analysts still treat a low or high HR/FB rate as fact that a pitcher suppresses homers or prone to them. Claims are made with no supporting evidence.
I think more generally that you are bothered by people mistaking correlation for causation, a story as old as time itself across any discipline
Yes that too. That’s simply not acceptable from a fantasy analyst.
At the same time, we shouldn’t completely dismiss an elevated HR/FB ratio. That’s why we regress instead of only using the league average.
Miguel Gonzalez has allowed something like a 12% HR/FB in a big sample. Let’s say league average is 10% for simplicity. When projecting Gonzalez I’m not going to ignore his history. I’ll end up with an expectation somewhere around 11% HR/FB.
Looking at the career HR/FB% figures for pitchers with 1000+ innings since 2002, Brett Myers has the worst total at 14.5% while throwing a fairly low percentage of fastballs, yet Brandon Webb is second worst at 13.2% while throwing a fairly high percentage of fastballs. There does appear to be some correlation between a low IFFB% and a high HR/FB%. If IFFBs are being included in the overall FB totals (some sites do and some don’t) then that explanation is obvious. Overall fly-ball percentage definitely matters, as there’s a noticeable correlation between low FB% and high HR/FB%. 24 of the 30 lowest fly-ball rates have a HR/FB% higher than 10% (Adam Wainwright sticks out as a pretty interesting outlier). Meanwhile in the other direction, 19 of the 30 highest fly-ball rates have a HR/FB% under 10%.
I also looked at heatmaps under the hypothesis that HR/FB% could be related to the percentage of pitches thrown in the middle of the strike zone, but noticed something else instead. What I saw, and what is presumably related to the above point regarding fly-ball percentages, is that pitchers who throw a high percentage of pitches in the bottom third of the strike zone have high HR/FB percentages whereas pitchers who throw a high percentage of pitches in the upper third of the strike zone have low HR/FB percentages. I found that amusing given the conventional wisdom that pitching up in the zone puts you at greater risk of home runs, which I suppose is still true in the aggregate but is the opposite of true on a percentage basis.
Clayton Kershaw – 18.4% upper, 22.5% middle, 15.8% bottom
Matt Cain – 17.4% upper, 22.7% middle, 15.8% bottom
Brett Myers – 13.7% upper, 23.5% middle, 18.8% bottom
Brandon Webb – 12.2% upper, 22.3% middle, 18.8% bottom
Now apply that to Brandon McCarthy:
TEX/OAK (7.36% HR/FB%) – 17.8% upper, 25.3% middle, 17.6% bottom
ARI/NYY (13.62% HR/FB%) – 13.4% upper, 25.1% middle, 21.3% bottom
When Brandon McCarthy threw up in the zone more than most, he had a HR/FB% better than most. When he changed to throwing down in the zone more than most, he had a HR/FB% worse than most. So what prize do I win?
Interesting. I think Dolemite and JDBolick give convincing explanations to explain why McCarthy’s HR/FB rate are elevated. Ground ball pitchers that miss their spots do give up more HRs.
Regarding JDBolick’s data, I wonder: has the emphasis on throwing low strikes created a generation of low ball hitters? I’ll go with a couple of hitters to try and support this hypothesis.
Last year, we saw that Mike Trout had trouble hitting high fastballs, especially high and tight.
Somewhere in your archives, you have video footage of every homer that Chris Davis hit during his 50 HR 2013 campaign. If you’re able to access that, you’ll see that the vast majority of those home runs were in the lower half of the plate. If you look at last year, you see that he had trouble with high pitches (and inside pitches as well).
Didn’t somebody on the FG staff publish a piece last year noting that the Oakland A’s had, prior to this year, assembled a group of low ball home run hitters that met with failure late last season when pitchers started throwing more high strikes against them—most notably Brandon Moss?
Also, Dave Cameron has complained repeatedly of how many low pitches in recent years have been called strikes.
It kind of reminds me of the mid 1980’s when American League umpires were instructed to shrink the strike zone. The upper part of the strike zone moved from the arm pits to just above the navel, with the lower part of the strike zone moving to below the knees. Hitters who were good at hitting low pitches thrived.
FWIW: my conclusion based on the arguments made above is that Brandon McCarthy needs to throw more high strikes. So part of it is bad luck, but part of it is his failure to respond to how batters have adjusted to a lower strike zone.
TY Baltic
Let’s not forget that it is Home run per fly ball
So there is a bit of a caveat that we must apply that high GB% give up less FB in general, therefore the ones that get put in the air get put in the air with authority.
I agree with you though. Based on Bollock’s numbers, and mccarthy has a good fastball velocity wise, if he threw a few more purposeful high ones, it is reasonable to surmise he might lower his HR/FB
Your theory regarding low ball hitters is plausible, but I also think it’s a matter of physics. I believe it’s more difficult to impart back spin on a ball high in the strike zone than one low in the strike zone due to swing planes.
Well only 4 players were cherry picked and correlation does not equal causation.
*sigh* Come on, Mike, be gracious. You posed a challenge and I answered it. Here are the numbers for the full top ten highest HR/FB% of pitchers with at least 500 innings since the start of 2007:
Brett Myers – 14.2 – 13.7 – 23.5 – 18.8
Hisashi Iwakuma – 13.7 – 11.4 – 22.9 – 23.7
Mike Leake – 13.6 10.3 – 22.5 – 22.4
Jerome Williams – 12.8 8.4 – 21.5 – 23.4
Braden Looper – 12.7 13.2 – 22.1 – 18.9
Armando Galarraga – 12.7 10.2 – 22.8 – 21.6
Tommy Hunter – 12.7 20.3 – 24.8 – 14.8
Chris Volstad – 12.6 11.7 – 21.8 – 20.6
Freddy Garcia – 12.5 14.0 – 21.4 – 17.8
Ricky Romero – 12.5 15.5 – 21.4 – 17.5
If you make me keep going I will, but the weird Tommy Hunter exception notwithstanding, the pattern could not be any more clear. And while correlation doesn’t necessarily equal causation, you need to come up with a compelling reason to explain why it does not in this particular case if you’re going to use that as an excuse to avoid admitting that you were wrong. You learned something today that will help your xHR predictions, Mike. That’s something you should be happy about, not a reason to be defensive and petty.
Those numbers are HR/FB% – upper third strike zone % – middle third strike zone % – bottom third strike zone %
This should be easier to read:
Brett Myers –
14.2 – 13.7 – 23.5 – 18.8
Hisashi Iwakuma –
13.7 – 11.4 – 22.9 – 23.7
Mike Leake –
13.6 – 10.3 – 22.5 – 22.4
Jerome Williams –
12.8 – 8.4 – 21.5 – 23.4
Braden Looper –
12.7 – 13.2 – 22.1 – 18.9
Armando Galarraga –
12.7 – 10.2 – 22.8 – 21.6
Tommy Hunter –
12.7 – 20.3 – 24.8 – 14.8
Chris Volstad –
12.6 – 11.7 – 21.8 – 20.6
Freddy Garcia –
12.5 – 14.0 – 21.4 – 17.8
Ricky Romero –
12.5 – 15.5 – 21.4 – 17.5
Thanks, I appreciate the work. However, I’d like to see a full correlation with all pitchers, not just a top 10 list. You also have to consider what several commenters mentioned about selection/survivor bias. A fly ball pitcher with a high HR/FB rate isn’t going to make it. But a groundballer could still get by.
Next, it could explain a higher than league average mark, but there’s still a line at which point something else is at play. You wouldn’t provide the same explanation if he had a 25% mark, right? So maybe the strike zone analysis pushes his mark up to 11%-12%. Now how about the rest of the way to his inflated mark?
http://www.sbnation.com/longform/2014/6/18/5818380/effective-velocity-pitching-theory-profile-perry-husband
In 2004, based largely on what many felt was a world-class fastball combination — a 97-mph four-seamer, a 91-mph cutter and a sinker that came in at 92 — Escobar signed a big-money contract with the Angels. Escobar managed to finish fourth in the league in strikeouts, but still posted a losing record for a club that won 92 games. Husband thinks he knows much of the reason why.
“Escobar’s stuff was about as good as it gets,” he said. “But at the time I did the study, the league was hitting his fastball combo at a .369 clip. If movement is everything for a pitcher, this guy should have been a world-beater. But if movement is so important, why was he getting killed?
“It turned out that he was throwing 97 down and away, which is about 92 EV. Then he throws the cutter to lefties at 92, and the sinker down and in, at about 91. He’s throwing all his pitches within 2 or 3 mph of each other [in EV terms], and he’s neutralizing all the effects. Even though the movement is there, it’s killing him. Guys are getting ready for one pitch at one speed, and receiving two bonus pitches at the same speed. He was throwing pitches that moved right into hitters’ bats, even if they were guessing wrong.”
Had Escobar reversed himself, he could have created a spread of up to 16 mph utilizing exactly the same pitches.
I agree that it would be great if I had a list of all pitchers, but I have to pull up the heatmap, select the appropriate years, switch to the 5×5 grid, and then add up the three figures manually. For the record, you do find the same thing at the other end of the scale where pitchers with low HR/FB% figures have a higher than average percentage of pitches up in the strike zone. Other factors undoubtedly matter as well, as most St. Louis starting pitchers join Hunter among those few exceptions who don’t fit the model.
Define luck. Skill is not the opposite of luck and vice versa.
For example, McCarthy could have thrown a lot of meatballs recently, but that might not be reflective of his true talent or skill at all, therefore it wouldn’t show up in the future. It wouldn’t have any year to year correlation. But it wasn’t luck.
But not every meatball is hit for a homer. If his HR/meatball percent is higher than average, that would constitute bad luck. And yes, I find it hilarious to have a discussion that involves meatball percentage. Mmmm, now I’m hungry.
One interesting thing I noticed is that the majority of home runs given up by McCarthy in the last year have come in the 2nd or 3rd pitch of an AB. Is it possible hitters are more aggressive against him and he has not adjusted?
Sure anything is possible. I’m looking for the evidence that your theory and anyone else’s claims are actually true.
Because he’s a sinkballer.
Pitchers who feature the sinker as their primary pitch typically have a high HR/FB%. Look at Edison Volquez for example—in 2014 his sinker had an HR/FB of 17.2%. Not very unlike McCarthy’s, which was 17.7% in 2014. Masterson was 15.2% in 2013 but inflated to 25% in 2014 due to a massive reduction in the velocity of his sinker.
Sinkballers let up more homeruns.
The question becomes, why does Kyle Lohse have such a historically-low HR/FB% on his sinker?
The answer: Movement.
Masterson’s average sinker movement is 9.3. McCarthy 9.8, Volquez 10. Lohse’s sinker movement averages 11.3, and last year it was 11.5—yielding an HR/FB of 12.9%. Still inflated versus a fastball/changeup guy, but less than the average sinkballer because of increased movement.
This is interesting, but I need more than 3 cherry picked numbers to be convinced that this is something. A full blown study with all pitchers would be more useful.
I would think that if you manage to hit a fly ball on a sinker, it is more likely to be crushed than a pitch up in the zone. Pitches middle and would seem easier to just get under the pitch and fly out. How often do guys “just get under” sinkers?
“Now, it has become trendy to acknowledge the luck component, but then side-step it and try justifying the out/underperformance.”
Thank you for saying this. It needed to be said.
The idea that groundball pitchers give up more homers is flatly wrong, especially once you account for the fact that they trend toward less strikeouts (strikeouts prevent homers cuz they’re not balls in play).
The idea that sinkerballers give up more homers is also necessarily wrong since sinkerballers get more grounders.
The idea that Brandon McCarthy in particular has a ‘hittable’ sinker is also wrong. Last year he threw 1558 sinkers and allowed 11 homers on them and threw 1467 of everything else and allowed 14 homers on those–and that’s with a far higher strikeout rate on the other pitches!
His career numbers also reflect this with an equal number of sinkers and fastballs and more homers on the fastball.
Also, guys with BB/9 rates under two don’t miss their spots.
So let’s drop those ideas. This is why Mike suggested finding actual evidence. It takes minutes to disprove those ideas.
I believe Rowen was talking about HR/FB, not just more homers, period. Touching on what was said earlier—“Let’s not forget that it is Home run per fly ball, so there is a bit of a caveat that we must apply that high GB% give up less FB in general, therefore the ones that get put in the air get put in the air with authority.”
You should re-read the initial task: To prove that McCarthy’s HR/FB% isn’t bad luck.
It is absolutely, statistically true that pitchers who feature the sinker as their primary pitch have a higher HR/FB% when compared to all other pitchers.
To further prove my statement, please refer here:
http://www.fangraphs.com/pitchfx.aspx?playerid=4662&position=P
As you can clearly see, the moment he began featuring the sinker as his primary pitch, his HR/FB% increased dramatically. Please don’t get upset. No one is saying McCarthy is an awful pitcher. The sinker does generate ground balls. But in the instance of justifying (statistically) why his HR/FB% is high—just look at historical HR/FB% in relation to his primary pitch.
As I stated to another commenter, correlation does not equal causation. A full study that looks at sinkers thrown by all pitchers would be enlightening.
has there been a pitch-by-pitch analysis that examines which are the most likely to go for home runs? it *seems* like a sinker that’s left up in the zone could be a very homer-able pitch but that’s intuition
I should follow this up with something else about McCarthy, albeit a bit off-topic but in light of this seeming like I am devaluing his GB% I feel it should be said. McCarthy looks great to SIERA because he generates a ton of ground balls. Which is true—and valuable to a team because the greater mission of a pitcher is to limit TOTAL runs…not HOME runs.
So, while his HR/FB% is much higher than that of a fastball pitcher (that would include McCarthy himself, when he used to feature the fastball as his primary pitch), he might still allow less total runs than a fastball pitcher. That is to say that if he featured the fastball as his primary pitch starting tomorrow, his HR/FB% would decrease dramatically. But he might allow more total runs.
I found no correlation between zone% or sinker usage and HR/FB.
I found a tiny correlation between o-swing and HR/fB R-Squared: 0.0524017.
http://www.fangraphs.com/community/what-is-more-important-for-a-fastball-velocity-location-or-movement-2/
suggests that maybe the location of the fastball is most important for this, so I’ll try that out.
Okay, but look, McCarthy allows a lot of homeruns, particularly for his profile so to say he is allowing more HR/FB because he’s a groundball pitcher does not explain why he has an inflated home run *total*.
An xHR/9 stat is going to have groundballers’ HR/FB profile baked into the formula. The one I built suggests his HR/9 rate should have been .75 in Arizona and .86 in New York so we’re trying to understand the more than .3 difference in his HR/9 profile rather than quibble over expected home run rates of groundball pitchers.
I couldn’t find a site that allowed me to see all the data from his HR’s given up, but I was able to watch the videos from my mlb app. All but one of his home runs were between 92-94 mph. I could not tell if it was his FB or Sincker though. The other homerun was on a hanging cureball in the middle of the zone.
A few of the HR’s barely cleared the fence.
Both of the HR’s to Cruz were low as away. Both went oppo To RF. Good pitches, but look who they were against! Seager and a Akley pulled theirs to RF. Wind was blowing out to RF at 3mph. Who knows if it was blowing more at times of hits though.
The only hard stuff pitch HR that wasn’t low and outside, was inside to Upton and pulled to left field.
Middlebrooks took him Oppo. Wind was 10 mph from left to right.
Seems like an unlikely number of opposite field home runs to me. Also, the overall quality of batters that hit them is pretty Damon good.
Would like to know if all of those hard pitches were sinkers or if they were Fastballs.
My head is spinning with all of this analysis. Bottom line — do I start McCarthy vs Col (Butler) on Sunday?
Ha! Yes, I will keep trotting him out there, especially in a home start against a team that is typically much worse in away games.
Unfortunately, Colorado hasn’t been home long enough for their swings to get all screwed up by the atmosphere. The Rockies are an easier road draw later in the season.
I’d like to note that if someone can at least show a correlation that’s pretty useful for the purposes of evaluating McCarthy’s fantasy stock.
Also, do they keep track of average flyball distance with pitchers the way they do with hitters? Seems like that might help tell something.
Indeed they do! I actually did an entire analysis earlier in the year and totally forgot about it. I’ll dig up the data for McCarthy.
With the exception of Kluber and Wainwright, I’d say this is a positive trend:
https://www.meta-chart.com/share/hrfb-rate-amongst-sinkball-pitchers
The more your throw the sinker, the higher your HR/FB% becomes.
very nice!!!
Piggy-backing off of this—there were only 6 pitchers who threw more than 40% sliders in 2014….four out of the six had HR/FB% way, way higher than league average.
http://i.imgur.com/YWIxGyG.jpg
Interesting, though above in the comments, Eno found no correlation between sinker usage and HR/FB rate.
Love the chart. Is that just for overall HR/FB, or is that for HR’s vs a sinker?
I mentioned that all but one of McCarthy’s HR’s this year have been in the 92-94 mph range by looking at the videos, but he throws his FB, cutter, and sinker all in those ranges. Tough for me to discern what pitches those were, but from looking at his pitch value/100 I’m guessing they were either sinkers or cutters, since he has poor values with those and great value with his FB so far.
*I’m wondering if it’s a problem that he throws 3 pitches all within 2 mph of each other…with only the curve that has velocity gap to use.
*also wondering if hitters are just looking down in the zone, and often away, as well as guessing a pitch of 92-94mph. Since 80% of his pitches are in the range. Seems like it might be easier for a big leaguer hitter to guess the time and location of his pitches maybe.
Seems like he might benefit from another off speed pitch.
I’m just eliminating things here:
One might posit that he is not hitting his spots *and* has a low walk rate because he throws straight down the middle every time, thereby delivering “meatballs.” The command vs. control thing. If this was the case, he would have a higher percentage of balls in the middle of the zone than other pitchers (since that’s what he’s aiming for).
However, compared to Clayton Kershaw, who had a 6.6 HR/FB last year and a virtually identical walk rate, McCarthy threw 9.1% to the middle of the zone vs. Kershaw’s 10.5% and I’d argue that no one thinks Kershaw is just throwing at the middle because he’s no better than that. Here are other comparables who have *low* home run rates:
Year Player BB/9 % to middle
2011 Fister 1.54 9.3
2013 Harvey 1.56 9.6
2014 Kuroda 1.58 7.3
2014 Alvarez 1.59 9.8
Sadly, I can’t aggregate that data so I can only count examples, but almost all low walk pitchers come in around 9. In fact, Kuroda, the exception here had a low HR rate for being in New York, but it was still the highest HR rate on the list.
And because I can’t resist, here is a guy with an abnpormally high HR rate…wait for it…
2012 Blanton 1.60 8.7
Yeah, the meatball factor is not a thing.
Gosh, good heavens, let me try to explain it one more time: a guy could throw 100% sinkers and get 90% ground balls and have the *highest* HR/FB rate in the league, but he still is going to have a *low* home run *total*. McCarthy is a guy with a *high* home run *total*. These are not the droids you’re looking for.
We are looking for what makes a home run total go up. Ground balls do not do that.
I agree. I’m making a dataset of 10 sinkballers and comparing them based on vSI, SI-X and SI-Z. There are some interesting correlations. I wasn’t able to draw any conclusions based on heatmapping though.
This site constantly extolls the virtues of the high fastball, something Mccarthy just doesnt use that often and probably should. It is entirely reasonable to conclude that if you throw a pitch over half the time, pitchers can sit on it. A few high heaters at 92-94 would do wonders for changing eye level, reducing predictability, etc…
http://www.fangraphs.com/fantasy/ryan-zimmerman-jacob-degrom-and-sitting-on-a-fastball/
“The best you can do against him is look for a distinguishable fastball and hope it’s of the sinking kind. So far, that’s what hitters look like they are doing. They are swinging at it 60% of the time as hitters continue focusing down. They have a .500 BA and .286 ISO against his Two-seamers relative to a .125 BA and .083 ISO against his Fourseamers. Ryan Zimmerman is our best example so far.
Hopefully, deGrom will pump up a few more Harvey-high Fastballs and outperform his 2014.”
There is some good discussion above and I believe there is some substance behind the argument that groundball pitchers will give up a higher HR/FB% than most flyball pitchers. Some have pointed out that this includes survivor bias – to which I would ask, so what? The question is whether some pitchers will consistently yield a higher percentage of home runs per fly ball, and even if this result is reached partially/mostly by survivor bias, the question is still answered and going forward it would seem plausible to more accurately predict HR/FB% based on the type of pitcher, rather than just applying the league average.
The only thing that I think should be added to this conversation is that in attempting to measure this “skill” I believe it would be better to determine average fly-ball distance rather than HR/FB%. Home runs are dependent on the park in which the game is played, and comparing the HR/FB% of someone that pitches at Fenway or Wrigley is not an apples-to-apples comparison to someone that pitches at Petco. Using average fly-ball distance would be a better metric to normalize any disparities created by park sizes.
Mike,
This has been an awesome discussion – it has been especially cool to see different writers and readers debating back and forth – and I’m glad it has inspired another one… would be awesome to see/read teams of fangraphs staffers debating more on murky subjects in the future.
Guys with high infield flyball rates tend to have lower hr/fb rates. Look at the data for 2005-2015. The higher IFFB% guys have below average hr/fb rates. Chris Young. Mariano Rivera, etc. Conversely guys with high hr/fb rates tend to have lower iffb% rates, although not necessarily the lowest rates. Guys going through period of high homers tend to be giving up flyballs and not pop-ups.
Also, we should not correlate “bad luck” with giving up a lot of long homer runs per flyball in a relatively short period. The fact that say, 11% of flyballs go for homers, and McCarthy himself may also give up about that many over time, just means he usually doesn’t pitch as badly nor do most pitchers who get to pitch for a long time in the majors.
Wobatus—well said. HR/FB% can be misleading at times, because it can make a pitcher look bad without context. It’s important to assess GB% and HR/9. There are good pitchers with high HR/FB%.
The key is that even guys who pound the bottom of the zone have to mix it up and throw some pitches up in the zone, and sinker/slider types still mix pitches without so much downward movement.
And for the non-superstar pitchers who use the above approach to get a high GB%, there’s probably a lot of selection bias that they’re below MLB average in their effectiveness with the higher/non-sinking stuff. So, for the relatively few times they allow a flyball, they stand a greater chance of it being hit hard.
I’m not a statistician, and I’m not certainly not discounting that luck has something to do with McCarthy’s ridiculous HR/FB rate, but I think his pitch selection has conspired to put him in this tenuous position.
Dataset: (All pitches in play in 2012 season)
PITCH TYPE %GB TOTAL BATTED BALLS
Splitter 50.30% 1607
Sinker 49.80% 12205
Curveball 47.50% 9985
Two-Seam FB 46.40% 17064
Changeup 44.50% 13481
Slider 41.00% 17735
Cutter 37.90% 6764
Four-Seam FB 35.20% 39782
McCarthy’s has made significant adjustments in his pitch selection (2013-2015).
Pitch Type 2013 2014 2015
Four seam 1.28% 13.15% 33.90%
Sinker 44.40% 48.68% 28.39%
Curve 14.72% 24.35% 22.88%
Cutter 34.75% 13.18% 14.83%
Pitches Tracked 95.15% 99.36% 100.00%
McCarthy has significantly increased the use of his four seam fastball which generates the lowest GB count among all pitches tracked. His four seam fastball is above average in generating fly balls when compared to other pitchers.
McCarthy has significantly reduced the use of his sinker in 2015 ( albeit a very small sample size) but this is his best chance for a ground ball in his repertoire. His sinker is also above average in generating fly balls when compared to other pitchers.
McCarthy velocity has gone up which should correlate with more ground balls. However, as the Fastball Times point out, changes in velocity and movement on pitches like four seam fastballs and cutters that are not good ground ball pitches do not produce more ground balls.
If you miss with the sinker bad things happen.
If you just think about avg. exit angle, pitchers that tend to induce groundballs have a lower average exit angle. this probably means that the flyballs they give up tend to have a lower exit angle, leading to more line drive type fly balls, hard hit fly balls, and fewer high fly balls and popups. Leading to a higher HR/FB, perhaps?