Choices

April 15, 2020
                                                                        Choices

            OK, recess is over; it is time to go back to working on the Runs Saved against Zero project. 

            Did you ever hear this joke?   Tom Johnson has not lived an honest life, so when he passes away the Devil claims him, and tells his right-hand man to show this guy around, have him choose his room.  So the Devil’s Assistant says, "Tom, you’ve got a choice of three rooms in which you will spend the rest of eternity.  You get to choose which room you want."

            In the first room there are loud obnoxious noises coming from all sides and bright flashing lights, and the room goes from brilliantly lit to dark, back and forth on a one-second pattern, and you can hear people screaming in pain, and everybody in the room is standing on their heads.   

            Then they go to the second room, and the floor is a heated grate; it’s like 140 degrees in the room and the floor is hotter, and everybody in that room is standing on their heads, too.

            So they go to the third room, and the entire floor of the room is covered with a foot of the most foul-smelling sewage you can imagine, and the room smells horrible and everybody in the room is covered with disgusting sores from living in sewage, but everybody is standing and drinking coffee.   So Tom says, "Well, it’s pretty awful in here, too, but at least you’re not standing on your head all day and you get coffee, so I guess I’ll take this one." 

            And he goes in, and the door shuts.  And the devil says, "OK, coffee break is over; all you guys go back to standing on your heads." 

 

 

            Anyway, back to work.  On what we are doing now I actually need your help.  I need your guidance, your input.  I need your feedback.

            We have to make a series of choices here, eleven of them.  We have 11 categories of data which contribute to runs being prevented.  While we might think simplistically that some of these are "pitching" and some are "fielding", the reality is that many of them, most of them, perhaps all of them except one, are responsibilities shared between pitchers and fielders.  

            We know, at this point. . . .well, we don’t "know".   We have formed an estimate, at this point, of how many runs each team has prevented by strikeouts, by control, by home run avoidance, by fielding range/DER, by turning the double play, etc.   What we need to do now is to take that information, and turn it into a binary split:  How many runs did this team prevent by pitching, and how many did they prevent by fielding?   One or the other; which is it? 

            There is probably some way to measure the fielder’s impact on each of these things and the pitcher’s impact on each of them, but we don’t have those measurements NOW.   Someone in the next generation will have to develop those methods.  For now, we’re just going to have to decide.   And your opinions on these things are probably as good as mine, so. . . .I need to know what you think.   Category by category:

 

            Strikeouts.   My first inclination would be to count the strikeout as 99% the work of the pitchers, and 1% the contribution of the catchers. 

            We’re working on the TEAM level here, dealing with TEAM strikeouts, not strikeouts by individual pitchers.   It seems to me that the catcher plays some minor role in this, not only by "framing", but also, at least historically, by calling pitches, knowing the hitters, setting up the hitters, etc. 

            I once talked about this with Joe Garagiola, who of course was a catcher; I asked whether he thought catchers had any impact on the pitcher’s ERA by calling pitches.   Joe said, "Nah, we don’t have anything to do with it," and he gave me this example.   He said that when he was with the Cardinals, he had Harry Brecheen and Howie Pollet and Al Brazle and Max Lanier; you’d get a hitter 1-2, and he’d call for a curve ball, and the pitcher knew that he meant a curve ball a foot out of the zone, make it look like it might be a strike and see if the batter would swing at it and get himself out.  But then he got traded to the Pirates, and he’d do the same thing, and the pitcher would throw a curve ball right over the plate.   He said he would think, "Jesus, you should have known I didn’t mean to throw THAT."  He specifically mentioned Murry Dickson as a guy who just didn’t get it, just didn’t have a clue what he was doing, although Murry was in his 30s by then, had been in the majors for a long time and somehow managed to win 20 games with a 7th-place Pirates team in 1951.   But his basic point was, you just can’t make the pitcher throw the pitch he is supposed to throw.  You just have to live with whatever he does.  

 

Walks.   Again, my inclination would be to count this as 99% pitching, 1% catching.   Maybe 2% catching for walks; the catcher wants to try to nibble on 2-2, it leads to a walk sometimes.  I don’t know.   I need your feedback. 

Home Runs Allowed.   Same thing.   99% pitching, 1% catching. 

 

Hit Batsmen.   Again, 99% pitching, 1% catching.

 

Wild Pitches.   Wild pitches are probably 80% the responsibility of the pitcher, 20% the responsibility of the catcher, maybe?  What do you think? 

 

Balks.  Who cares?   There are so few runs involved here it doesn’t really seem to matter.   Just place the responsibility all on the pitcher.

 

DER.   This is really the monster item on the day’s agenda.  Making the right choice here is vital to the success of the project—and there is a huge array of credible options.  

            We all know, from the work stemming from Voros’ original publication, that the pitcher has relatively little impact on whether the balls put into play against him become hits or outs.   The balls put into play against Gerrit Cole have more or less the same generalized outcomes as the balls put into play against David Hess; once the ball is in play, it’s in play. 

            The pitcher has relatively little control of that, but not zero control of it.  But does the pitcher get 5% of the credit here, 20%, 2%, what?   I could make an argument. . . let’s call this argument A.

            Argument A.   I could make an argument that the pitcher gets 70% of the credit for outs on balls in play.   The argument is, 70% of balls in play result in outs.   The pitcher, when he gets the ball put in play, has succeeded 70% of the time.   He hasn’t walked the batter; he hasn’t let it him hit the cheap seats.  He hasn’t gotten him 100% out, like he has on a strikeout, but he has gotten him 70% out. 

            The response to that is Argument B.

            Argument B.    The pitcher has already been given credit for putting the ball in play.  There are five outcomes:  the plate appearance can result in a strikeout, a walk, a home run, a hit batsman, or a ball in play.   We’ve accounted for all of the things that the pitcher does—the strikeout, the walk, avoiding the home run, avoiding the hit batsman.   This is what’s left.  This is the fielder’s little share of the pie.  You can’t give THAT to the fielder, too, can you?

            I’m 80% sure that Argument B is entirely correct, and Argument A is just flat-out wrong.  But I’ll need to look at the data once I have it organized, and I want to hear what you all have to say about it.

 

            Fielding Percentage:   Appears to be 100% the responsibility of the fielders, right?   I mean, the pitcher is involved as a fielder, but the pitcher is not involved in the allocation of responsibility for his role as a pitcher.

 

            Double Plays:   Probably 90% the responsibility of the Fielders, I would say?  The pitcher has SOME responsibility; some pitchers are willing to put the guy on first if they think they can get a ground ball; some are not going to do that, and would be silly to try because ground balls aren’t their thing.  

 

            Stolen Base Value:   Maybe 2/3 the responsibility of the catcher, 1/3 the responsibility of the pitchers?  

            On the individual level, the pitcher’s role in preventing the stolen base may well be larger than the catcher’s.   On the team level, teams are a mix of pitchers, and a mix of players generally is somewhere near the center; thus, on the team level, it’s mostly the catchers who determine where this team stands compared to other teams.  So I am kind of thinking 2 to 1.

 

            Passed Balls.   Again, about 2/3 or 70% the responsibility of the catchers, 1/3 or 30% the responsibility of the pitchers, I think. 

 

            In addition to that, I would value your input into one collateral issue.  When I am studying this issue, working out the system in more detail, I will need some "sample teams" to uses for study/illustration.  I would value your input into what teams I might use.

            Thanks for reading, and thanks again for your input. 

 

 

 

 

           

 

           

 

 

 

 

 

 

 
 

COMMENTS (28 Comments, most recent shown first)

MidnighttheCat
Thinking through my comment, about the role of the pitcher in DER, looking at the 1969 Orioles (just to pick a team known for a certain level of quality defense behind the pitchers), Batting Avg. on Balls in Play was .230 for Jim Palmer AND for Mike Cuellar, which absolutely suggests that it is the defense an not the pitcher that matters, but it was .262 for Dave McNally. What would be a statistically significant difference on pitchers who had the same defense behind them pitching half their games in the same park and very often pitching the same 3-game series in other parks? And why is it lower for Hardin and Pheobus, the other starters who were inferior pitchers, .251 and .257, than for McNally?

In 1970, it was .262 for Cuellar, AND also .262 for Palmer (weird no? two years in a row) but was .274 for McNally, was .234 for Pheobus but .289 for Hardin.

Why is McNally's number so much higher than for Palmer and Cuellar two years in a row if he is having no real effect on the result and the fielders are almost always the same ones for each of the three starters. And why the other differences for the other two?

Looking at a very different, much more asymmetrical example, the 1972 Phillies, the numbers for the other pitchers with a fair number of starts are: Reynolds .277, Fryman .298, Brandon .302, Twitchell .316, Champion .317, but for Carlton it is .259.

Now if Steve Carlton's ONLY contribution is HRs, Walks and Ks, why is his Batting avg. on balls in play (BABIP) against him so much lower than his teammates in the rotation who often would have the same defense playing behind them in the same ballparks? If we are to attribute Steve Carlton's 1972 record to his good luck, that WOULD be innovative !! And it would not really pass the common sense test I think.

NOW, weirdly, there were two other pitchers with a number of games started with LOWER BABIP than even Carlton on that team: Selma at .257 and Lersch at .245, and the case of Selma who seems to have been pretty awful that year seems to pretty strongly instead suggest that I am on the wrong track here. He walked a lot of people and gave up a lot of home runs per 9 innings, hence his very high ERA, despite the great BABIP stat. And he struck out fewer than he walked. So score a point for the three pure outcomes. Lersch though had a pretty good year.

If we were to consistently find two things: 1) pitchers of relatively similar quality on the same team having different BABIPs season after season, and 2) better pitchers (as measured otherwise) giving up fewer hits on batted balls than less quality pitchers, with their own teammates as a control group, this doesn't mean pitchers are having that big an effect on things other than Ks, BBs and HRs, but it does warrant further investigation before we exile pitchers to the outer darkness.
6:41 PM Apr 19th
 
MidnighttheCat
You won't love my "guidance" here, but here goes anyway in the interests of intellectual honesty, my learning something from you showing me I am wrong etc.

So I have never been convinced by the idea that the pitcher has almost no role in what happens to batted balls.

I think what is happening here is a meta-methodological problem. What I mean is that because TREATING batted balls as the defense's responsibility makes it EASIER TO MEASURE, that the lure, the seduction of THINKING THAT THAT IS AN ACTUAL FACT is having an effect on the effort to study the problem.

After all, what if the pitcher does have an impact on what happens to batted balls? Even a big impact? That would make the problem harder to measure, and to analyze.

Yet nearly every pitcher I ever remember seeing comment on this question, such as Sandy Koufax saying he had become a good pitcher by not trying to make the batter miss the ball but rather to hit it, sees this issue differently. Granted it is in their own interest to do so. But no more than it is in the interest of analysts to have something they CAN analyze instead of something they cannot.

Hitting a ball dead-on (and so getting more "exit velo..") surely makes a ball less likely to be fielded than hitting some corner of it and nudging it toward a fielder. Is it really believable that pitchers have not influence on whether batters hit their pitches well or badly? That the only thing pitchers do is the "three pure outcomes?".

Is there a study of exit velocity by batters off of different individual pitchers? Wouldn't that kind of data help us figure out if some pitchers are better at not getting hit that well when they do get hit, getting us past the Holy Trinity of walks, HRs and Ks?

To me, DER is 50% pitcher, 50% fielding until data shows differently. "Common sense" or "what seems logical" is not good enough here when flying in the face of what pitchers themselves seem to think they are doing out there. Not that they can't be wrong, but until data is convincing, I see no reason to err on the side of assuming that pitchers' role with respect to batted balls is equivalent to how the fashion industry sees models as "clothes hangers" and the clothes as the real story.


6:07 PM Apr 19th
 
CharlesSaeger
When was splitting things 50/50 giving a majority of credit to the fielders? Also, 33% of the variation in team BABIP is random chance, and most of that 33% is in the 86% you're citing. (And, you're the one bringing up xBA in the first place, not I; I've hitherto ignored its existence until this discussion.)

I'm going to pause after writing this graf and take a look at the Statcast data for out distribution by platoon percentage. I haven't seen evidence that the pitcher can influence the lateral distribution of balls in play beyond what the force and angle of the ball off the bat is. I'm going to look at the platoon data over the last few years because that's one place where it should absolutely show up if there is any ability. Going into this, I'm going to say I don't expect much, mostly because the platoon differential is the three true outcomes, though there is some, and furthermore, pitchers do vary widely in platoon differential.

OK, back. There might be some ability here. It's faint in the data, and clouded by lefty pitchers being more likely to be groundballers. It shows up mostly in the outfielder numbers. I assume infielder differences mostly come from positioning. I used all balls hit 2016-2019; next I move to just ground balls, to get away from the angle/velocity issues. Furthermore, this will make the outfielder data much more meaningful, as if an outfielder fields it, it's a hit.

Now ground balls only wasn't more helpful, though again it looks like there might be something. The biggest change for outfielders is that against righty pitchers, lefty batters hit 1.6% fewer ground balls to right field and righty batters hit 1.0% fewer ground balls to left. For infielders, lefty batters are less likely to pull grounders against lefty pitchers, though righty batters have no change.

Now, I'm going to use just right-handed batters, exit velocity between 85 and 90 mph, launch angle between 10° and 25°, giving 4,104 balls in play between 2017 and 2019. (It came to slightly less, I presume due to data not being coded. I had expected slightly more due to overlapping angles, as in -45° to -30° and -30° to -15° sharing the -30° angle.) This is the one I've wanted to see for awhile, actually.

I can't see it, and this is where I'd expect it. Against right-handed pitchers, right-handed batters hit 6% of all balls (with a launch angle between 10° and 25°, exit velocity between 85 mph and 90 mph) to a lateral angle between -45° and -30° (third base), 24% between -30° and -15° (3b/ss hole), 26% between -15° and 0° (shortstop), 25% between 0° and 15° (second base), 15° and 30° (1b/2b hole), 2% between 30° and 45° (first base). Against left-handed pitchers, right-handed batters hit 6% to the third base segment, 24% to the 3b/ss hole, 28% to the shortstop segment, 24% to the second base second, 16% to the 1b/2b hole, and 2% to the first base segment. Or, against righties, they pulled 56% of these balls, and 58% against lefties. That's barely anything; most of the difference is to shortstop, and that might be some coding quirk. Looking at that angle another way, righties hit 50% of balls up the middle against righties, and 51% against lefties.

I'd expect to be better able to predict how well a fielder handles the ball knowing the lateral angle; I don't think anyone would think otherwise. I just can't see how the pitcher has much to do with that, beyond what he already does from controlling the launch angle and the exit velocity. Those do have influence on the lateral angle; ground balls and line drives are pulled, while fly balls and popups go the opposite way. But when I try to control for that and substitute the most unbiased way of handling pitcher quality—the platoon advantage—I get no change.
6:33 PM Apr 18th
 
Guy123
Again, Charles, xBA is only a (relatively) crude measure of how likely each BIP is to become a hit, capturing velocity and launch angle. A hard hit grounder to the first baseman is counted the same as one hit to the SS-3B hole. A line drive to the gap is equal to one hit directly at the CF. The Statcast data can tell us much more: how close or far each ball was to the nearest fielder. And using that data, we find that the characteristics of the ball off the bat explains the large majority of variation -- 86% -- in team DER.

(This 86% also captures any ability that teams may have to position their defenders better or worse than average. To the extent this does vary, it's not clear we want to credit the fielders for it since it's the coaches' contribution. But even if you do, it can't possibly be a very large factor given how similar defensive alignments in baseball are.)

In other words, if we literally knew nothing about the fielding ability of each team's position players, we could still explain the large majority of DER variation. I don't see how we can square that fact with giving fielders most of the credit.
1:59 PM Apr 18th
 
CharlesSaeger
Oh, I agree that 80-90% is extreme, which is why I was astonished. I looked at those data another way just now, and came up with a more satisfying answer:

BABIP for 2019 has a SD of 0.0126. Again, xBA (with HR and K removed) has a SD of 0.0086, so that's 46% of the variation.That definitely includes pitching.

HR isn't negligible anymore. The SD of HR allowed last year is 28, ranging from 181 (Tampa Bay) to 305 (Baltimore). They're 16% of all hits. That's a big deal.

Do pitchers show variance on how often batters pull balls of a certain launch angle and exit velocity? I know the pitcher can influence the angle and velocity, but those lead to certain kinds of drives, and there are patterns they display. Batters have a strong tendency to pull ground balls, and the rate doesn't vary that much among them; popups tend to go the opposite way. Line drives are usually pulled.

What else is included in that 46%. I assume that ballpark is not included in xBA from the definition on MLB.com, so that's in the remaining 54%. Is luck also included? I presume it is, but part of the whole reason for xBA is to get rid of some of the luck.

Best number I can say from all this is 50/50. Which seems like a copout.
1:58 PM Apr 17th
 
Guy123
I downloaded the xBA numbers from Statcast, then plugged in the AB, H, HR, and SO so I could generate an expected number of hits, then reverse-engineer an expected BABIP and its SD is 0.0086. The SD of the difference was 0.0083, so this implies the fielders are responsible for 95% of BABIP differences

Charles: I'm not sure what you mean by the "difference" here. In any case, my estimate was based on xBA excluding Ks. So HR are included, but I don't think that would change the SD much. Also, xBA certainly underestimates pitchers' influence on the out likelihood of each BIP, as it doesn't consider the direction of the ball (a line drive to the gap and one hit directly to the RF are rated the same). If we also consider proximity to the fielder (by reverse-engineering OAA), the SD rises to .011, nearly as high as DER itself.

I think it's undeniable that the characteristics of a batted ball *off the bat* explains the vast majority of outcomes. The question then is, what does that mean for allocating credit/blame? My inclination is to give most of the responsibility to the pitcher, who after all threw the pitch and allowed the resulting contact.

BUT, that does credit pitchers with outcomes we know include a lot of random variation (luck). So an equally valid approach is to divide responsibility based on our best assessment of the variation in *true talent* that pitchers and fielders respectively have to prevent hits on BIP. As you know, that's a very complex issue. I think something like a 50-50 split is likely there. Many great pitchers did have a substantial ability to suppress BIP -- I remember calculating once that Seaver (and many others) owed 30% or more of his success to avoiding hits. (You have to look at career stats to discern this.)

I don't see any plausible argument for assigning 80-90% of the responsibility to fielders. There is simply no evidence that the fielding ability of teams varies that much.
7:28 AM Apr 17th
 
stublues
Re stolen bases. Is it worth giving some credit to the infielder for catching the throw and applying the tag?
7:11 AM Apr 17th
 
shthar
So a TEAM gives up a passed ball?

hmmmm. Interesting.

Gee, I guess Hal was crazy trying to limit the determining factor to pitchers & catchers.






10:54 PM Apr 16th
 
CharlesSaeger
I downloaded the xBA numbers from Statcast, then plugged in the AB, H, HR, and SO so I could generate an expected number of hits, then reverse-engineer an expected BABIP and its SD is 0.0086. The SD of the difference was 0.0083, so this implies the fielders are responsible for 95% of BABIP differences.

That's higher than even I expected. So maybe I'm missing something. Maybe the expected doesn't adjust for park and includes luck.

My expectations come from the Solving DIPS article from Tom Tango and Rob Wood from 2003. They come up with a split for an individual pitcher for a season (700 BIP, which seems astronomical now), and pitcher BABIP is broken down 44% Luck/28% Pitching/17% Fielding/11% Park. For a team, the individual pitchers are going to going to converge and luck is going to even out, so I'd expect Fielding to overtake Pitching.
9:00 PM Apr 16th
 
CharlesSaeger
Guy123: Interesting, but I'm pretty sure that xBA includes HR and K. (I also get a higher SD for it, 0.014. The SD of expected vs. different is 0.006, which is in line with your figure for OAA.) I'm going to take a closer look at this one.
8:12 PM Apr 16th
 
CharlesSaeger
shthar: Hal Richman was in the business of rating players, not teams, and was openly giving Gold Glove winners automatic top ratings anyways
7:52 PM Apr 16th
 
Zeth
On DER: I, too, am confident Argument B is correct. If you look at the gap between the Batting Average on Balls In Play allowed by the best pitchers at limiting hits on balls in play versus the average, long term, I believe it suggests fielders should get at least 80% of the credit for team-to-team variance in DER. Maybe as high as 90%.

On Wild Pitches: When I charted games for Baseball Info Solutions years ago, we always tracked Catcher Blocks; a Catcher Block was any ball in the dirt that didn't become a wild pitch. I intuitively believe a nimble catcher must block more pitches in the dirt and pick them up quickly, thus prevent them from becoming Wild Pitches, than a lumbering catcher. Therefore I am inclined to give catchers 5-10% of the credit for preventing wild pitches. The data could help us here: on a team that, in a given season, had two catchers see significant time, how often do the pitchers get charged with more Wild Pitches with one behind the plate than the other? I would guess you would find a small but persistent difference, but I don't know.

On Credit for Strikeouts/Walks/Home Runs Allowed: For years people tracked cERA, catchers' ERA, believing some catchers helped pitchers pitch better, and for years after that fell out of fashion people have tried more complex ways of measuring Pitch Framing and the like. My feeling about this is: if you believe that, in hard-to-measure but nevertheless real ways, a good catcher helps the pitcher allow fewer runs than an average catcher, where else would that show up, if not in the Strikeouts, Walks, and Home Runs? Do we expect that effect (if it exists) to show up only in the form of DER, results on balls in play?
3:21 PM Apr 16th
 
Guy123
On DER, I think you need to give at least half the credit to the pitchers.

If we look at team Outs Above Average (Statcast's fielding metric), the SD for outs on BIP was just .004, or about 18 runs. Statcast also has a metric, "expected batting average (xBA)," that estimates the likely batting average allowed for each pitching staff based only on the characteristics of the ball off the bat (exit velocity and ball angle). The SD for this metric last year was 0.008, twice as large as the fielding SD. Even if you want to give fielders a lot of credit for the quality of their positioning (not measured by OAA), it seems clear that pitchers have at least half of the responsibility for DER. In fact, it's probably more like 60-75%.

This is not surprising when you consider the distribution of balls in play: the large majority are either an easy out or an obvious hit, so fielders have little impact (and when they do, it's mostly in the form of errors and captured by fielding %). Now, it's true that preventing hits on BIP is not a *repeatable skill* to a very large degree. Thus, Pedro could post a .325 BABIP/DER in 1999 and then .237 in 2000. But little of that change can be explained by Boston fielders -- Pedro just happened to give up BIP that were much harder to field in 1999.
I understand the reluctance to "blame" Pedro for that, but it makes even less sense to blame his fielders. If we had Statcast data from 1999, it would surely show a high expected BA, based only on how hard the balls were hit. How can we hold fielders responsible for that?​
3:17 PM Apr 16th
 
shthar
You know, I think Hal Richman figured all this out a while ago.

I'd use his numbers.


10:45 AM Apr 16th
 
willibphx
As far as teams to analyze, I would suggest the Orioles from 68-72. Few reasons, these teams are generally considered to have been very good at preventing runs but lots of debate between how much was pitching versus fielding. Same park across all years and extremely stable rosters defensively and pitching. 10 pitchers accounted for 85% of the batters faced over the five years. There were only 12 primary starters for the 8 positions during that time as well.

May limit the full analysis across eras, high and low run environments etc but might provide some level of validation.
10:26 AM Apr 16th
 
jstodola
I think wild pitches are 99% pitcher. The catcher really doesn't have any control over where the pitch ends up.
9:38 AM Apr 16th
 
CharlesSaeger
TL;DR of what I wrote: We can study DER to figure out the pitching staff's influence on it and come up with an educated guess of what to credit to the pitchers rather than throwing out numbers based on our personal observations about weak contact and hitting behind in the count.
9:22 AM Apr 16th
 
CharlesSaeger
I'm going to limit things here to DER.

Pitchers do have some control over DER and should get some credit for this. Moreover, it should be possible to make estimates of how much credit pitchers get for this based on other characteristics of the team. This is something like what Bill already does with DPs: teams with more assists have more DPs, so you adjust for this. (Aside: if you want to actually give some credit for the pitchers for DPs, you need to be using a number that doesn't adjust for assist rate.)

The real issue with splitting responsibility for DER is that we have only a limited understanding of these biases, since they're secondary. Take strikeouts. I'm pretty sure that teams with high K rates have higher DERs, but strikeouts themselves do not have a direct influence on it since we've taken them out of the equation. It's more of a tendency, and the exact level of this tendency is a matter of debate. Same story with assists: I'm pretty sure that high assist rates, even after adjusting for strikeouts, indicate a lower DER, even though the assists themselves are counted as outs in the DER equation. Voros's second version of DIPS took some of this into account, albeit for individual pitchers.

It is possible to study this issue given a decent data set. Basically, sort all teams by expected DER versus actual, and see what are the other statistics of the teams that had a much better DER than expected versus those that had a much worse DER than expected. Do the high DER teams have low assist rates? If so, then you start adjusting DER for assist rate. Do they have high walk rates or low walk rates? Same story.

After these adjustments, almost all of what's left should be given to the fielders. Maybe a little goes to the pitchers, but most goes to the fielders. You do need to have the unadjusted number to allocate to the pitching staff as a whole.
9:17 AM Apr 16th
 
jfenimore
Fielding percentage depends a bit on the pitcher, I think. Some pitchers keep hitters off balance frequently enough to get more routine ground balls than those pitchers who give up lots of rockets off the bat.
8:51 AM Apr 16th
 
jfenimore
Walks are 99% on the pitcher. It doesn't matter if the catcher wants to nibble. The ball is in the pitcher's hand.
8:49 AM Apr 16th
 
Brian
My first instinct based on some of the pitch framing data from recent years is that 1% is too low for catchers for strikeouts, walks and homeruns.

While we only have data on that for a relatively short time in baseball history, I think the best guess that pitch framing was even a bigger part of the game in the years before we had data. This is because:

1)With their bulky equipment umpires wore in the past is it fair to assume that the job of calling balls and strikes was more difficult? Also 2 periods at least would seem to have visibility issues for umpires. The era prior to the banning of the spitball and the era where night baseball was becoming more prevalent, but the lighting wasn't as good as it is now. Also, just the general point that I would presume that umpires, like everybody else, have improved over time.

In my view, anything that made it more difficult to call balls and strikes would lead to more reliance on cues from the catcher.

2) To the extent that catchers were successful at doing this, there was no TV strike zone to monitor the performance individual umpires. Likewise, because no pitch framing data existed for catchers, umpires were not put on notice that catchers were successfully tricking them into calling balls and strikes.

Anyway, since your past research show that even changing ball and strike calls just a little has an exponential effect on the success of the plate appearance, I would think that the catcher's ability or to change the call is probably worth more than 1% responsibilty of the "three true outcomes" plate appearances.
8:00 AM Apr 16th
 
Brian
To clarify, do you want our input on the choices you are making in assigning runs prevented responsibility or which room we would choose for Tom Johnson?
7:37 AM Apr 16th
 
ajmilner
Yes, I first heard the Tom Johnson joke in middle school. And it wasn't sewage they were standing in.​
7:26 AM Apr 16th
 
studes
By the way, my comment applies on both a team level and individual level, FWIW.
5:17 AM Apr 16th
 
studes
I’d like to speak up a bit for Argument A. If a pitcher gets a batter to hit an infield pop up (as the most extreme example), I think he should get 100% credit for the out and no debit if the ball isn’t caught. He’s done his job. And you could make a similar argument for many other types of “easy outs.”

Conversely, if a ball lands in no-man’s land as a hit, that should be 100% on the pitcher. Everything in between the pop up and no mans land gets shared as credit.

Even if you take the strong version of Voros’ finding, (and I think pitchers do have *some* impact on batted balls), I don’t quite buy that pitchers shouldn’t be included in the calculus of responsibility. Outcomes are outcomes and some balls are clearly easier to field than others.

If you only have DER and nothing else, you would need to look at other variables to guesstimate the relative difference in allocation. For instance, home runs are probably given up by flyball pitchers and flyball pitchers give up more infield flies. That sort of thing. I’d start at something like 60/40 and adjust from there.
5:11 AM Apr 16th
 
Dhandforth
As for teams, perhaps the most fruitful will be teams that lots of 20 game winners or pitchers with sub 3.5 ERAs. Unless strikeouts dominate (1967 St Louis Cardinals), those sorts of teams tend to have a lot of pitching/fielding plays to study. So suggest 1968 Detroit Tigers, 1973 Oakland A's, and of course, 1997 Atlanta Braves.

2:58 AM Apr 16th
 
Dhandforth
Hi Bill:

Great study here. Am following your process with anticipation. Thanks for asking. Here’s my take:

Strikeouts: Agreed

Walks: I couldn’t find the data on IBB as a % of BB. But I’d have to say that IBB’s are 50/50 pitcher and catcher. And that’s going to change the ratio to something like 80/20 (give me a few hour and I can do the math).

Home runs allowed, Hit Batsmen Fielding Percentage, Double plays, Stolen Base Values, Passed balls: agreed.

Wild pitches: that’s a tough one. I once had (the good fortune of having) dinner with Tim Wakefield and asked him about his 134 wild pitches. And he said that even given the best intentions and best preparations between the pitcher and catcher, sometimes those balls have minds of their own. 80/20 is a fair guess. But I think there’s also an argument for 50/50.

Balks: I don’t care either.

DER: with respect to Hortonwho’s comment: don’t see the difference between hard hit balls and soft hit balls. If the implication is that hard hit balls produce more hits, then I’d like to have a discussion on what is a “hard hit ball” and then do a study between hard hit line drives that are quick outs vs soft hit ground balls (and soft hit bloopers) that turn into hits.

As for Alternative A vs Alternative B: Agree that Argument A is wrong. For Alternative B: For balls put into play, they are either fly or ground ball outs, fly or ground ball hits, or errors. So the errors are clearly on the Fielders. And I’d argue the outs are on the Fielders. That leaves hits: we’ve accounted for Home runs. Hits to the outfield, doubles down the line and against the wall, should also be on the Pitcher. Versus hits where the fielders are shifted or guarding the lines. Or doubles that are stopped as singles because of a good fielding throw. Or botched double plays where a runner is safe or almost any other nuance for balls put into play that result in a runner. Those are on the fielders. So even for Alternative B, there is something on the Pitcher. Lazily waving my hands in the air, can we call that 30% on the pitcher?

James
2:45 AM Apr 16th
 
hortonwho
Isn't slugging average on balls in play important? I mean -- there are kind of 6 outcomes, right? Strikeout, walk, homerun, hbp, hard hit ball in play, weakly hit ball in play? What I mean is, even if the pitcher has little impact on whether or not a ball in play is a hit, does he have some impact on whether or not it's an extra base hit?

Apologies if research has resolved that question, but it seems to me that there might be a greater divergence in hard bit balls in play vs weakly hit balls in play based on the pitcher's ability?
10:06 PM Apr 15th
 
 
©2020 Be Jolly, Inc. All Rights Reserved.|Web site design and development by Americaneagle.com|Terms & Conditions|Privacy Policy