Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Friday, August 24, 2018

Posnanski: Baseball 100 Rules

In this era of reboots, it was perhaps inevitable that Joe Posnanski would take another crack at the 100 greatest players in major league history. 

The Baseball 100 is more than just a ranking system to me. The difference between my 78th ranked player and my 212th ranked player is so miniscule that it’s mathematically irrelevant. With one slight adjustment, I could have those two players switch places.

Nearly all of the series is to be pay walled, but Zach Greinke is No. 100 on the list.

In the original version of this list, I included a bunch of Negro leaguers — I can tell you that four were in my Top 20. I still believe this. But Negro leaguers will now be a major part of my corresponding Shadowball 100….It’s an eclectic list that includes players who are, in their own ways, larger than life.

No. 100 on this list is Duane Kuiper.

 

 

Rennie's Tenet Posted: August 24, 2018 at 08:01 AM | 282 comment(s) Login to Bookmark
  Tags: history, joe posnanski, joe posnanski top 100, reboots

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

Page 2 of 3 pages  < 1 2 3 > 
   101. SandyRiver Posted: August 27, 2018 at 03:26 PM (#5734310)
We do have 146 PA against Lefty Grove. Ruth’s strikeouts were way up against Lefty’s heat, 45 in 133 AB, slightly more than 33%. He was still very productive against Lefty (316/377/534) with 9 homers.

Babe's OPS+ 1925-on (Grove's 1st year) was 200. Up thru 1924 it was 218, so even those respectable numbers might be an underestimate. Add the 9% OPS+ difference and they're 344/411/582. (And maybe another dinger.)

That's not bad, especially considering Grove had the lefty-lefty advantage working for him.

Not much there. Ruth's splits show OPS 1.152 against RHP, 1.128 against lefties. They're not complete, but one would assume more from Grove's time than before that.

I'm probably just obtuse, but I can't get the UER "discount" on a hitter's value. Consider 2 scenarios:
#1
Batter A doubles.
Batter B plates him with a single.
#2
Batter A doubles.
Batter B grounds toward 2B, the fielder lets it go thru his legs, runner scores.

Obviously, batter B gets no credit, but why would batter A's double be worth less in #2?
   102. McCoy Posted: August 27, 2018 at 04:03 PM (#5734348)
I've piqued my curiosity on lefty usage during Babe's playing days. I have a working theory that managers did not use left handed pitchers optimally back in the day. The first issue is that managers back in the days did not use relievers properly. They had worse numbers than starters. They didn't really understand platoon effect either. Then on top of that (and this is my theory) managers left their pitchers in too long so while Ruth, Cobb, Jackson, et al would have to face a lefty starter they also got to face a tiring lefty starter in the mid to late innings as well. Thus unlike modern lefties who generally have the vast majority of their LHP PA be one or two PA those guys from the old days got to feast on pitchers in the 7th or 8th who shouldn't have been in the game.

I did a look on BRef, but what to do a more detailed look with retrosheet, at Ruth's 1925 season and he does in fact have his worst showing in the first PA and gets substantially better the 2nd, 3rd, 4th, and the 5th time facing a pitcher. In fact he has something like 45 odd 1st time PA against lefties and they don't give an extra base to him at all. Now imagine if instead of facing a tiring lefty in the 8th or some below mediocre right handed reliever in the 8th he faced a LOOGY at least once every single game.
   103. Mefisto Posted: August 27, 2018 at 04:42 PM (#5734377)
Obviously, batter B gets no credit, but why would batter A's double be worth less in #2?


Consider 2 scenarios:

1. In #1, fielders never make errors. In this scenario, A gets stranded on 2B and his double will never lead to a run because B will ground out.

2. In #2, fielders make lots of errors. In this scenario, A will score X% of the time not because B did anything, nor because A did anything different, but due solely to something the defender did. Because A's double leads to more runs in an environment with more errors, A's double has a higher coefficient in a linear weights formula which purports to measure offense, even though the offense did nothing at all different than it did in scenario #1.

   104. McCoy Posted: August 27, 2018 at 06:19 PM (#5734463)
Ruth had a .559 OPS against lefties when facing them for the first time in a game in 1925. Overall he was .775 and .825 against starters and relievers respectively when facing them for the first time.
   105. Sunday silence Posted: August 27, 2018 at 08:14 PM (#5734513)
As I believe Walt mentioned Ruth has the highest on contact numbers in the history of the game for a player with more than a blip of a career. It would be hard to believe that Ruth would significantly improve on those number


this is the lynch pin of the entire reasoning. both yours and walts. and it makes no sense. consider these sentences;

(let's say its year 2000 to exclude barry bonds for a moment)

Ruth's HR rate was higher than anyone in history, therefore it would be extremely hard to believe he could improve upon that.

Does that make sense?

What about this one:

Ruth's strike out rate was higher than anyone in history (just assume that fact) therefore its very hard to believe it could get any worse.

You seem to accept that the second sentence cannot be true and yet you fight tooth and nail to deny the first sentence cannot be equally true by the same reasoning.

We could play this game a lot. "Bob Feller had the highest strike out rate in history, therefore its impossible to imagine him improving upon that.' yea or nay for that one?

Or how about Ty Cobb and stolen bases? What about Bill Mazeroski's fielding numbers? they can't go up in a different era?


I dont get it.
   106. Sunday silence Posted: August 27, 2018 at 08:22 PM (#5734516)
Because A's double leads to more runs in an environment with more errors, A's double has a higher coefficient in a linear weights formula which purports to measure offense, even though the offense did nothing at all different than it did in scenario #1.


Lets restate that with a different premise. Assume the runs per game is the same in both cases. say its 4.5 runs/game in both eras.

Does the double weight any differently?

It seems like you might have to make an adjustment for players who put the ball in play. Because as you suggest you can score more with a ball in play in scenario 2. But if we're talking about the dead ball era, there arent many high KO batters so probably this adjustment doesnt make much difference. Not saying you dont have a point here but in terms of pre 1920 I dont think it can possibly impact Runs Created measurement very much.

If we're comparing say the 1920s lively ball era with modern times, isnt the run environment about the same? Somewhere around 4.5/ game? And if so then there would be nothing wrong with weighting doubles the same in both eras. Correct?

As for eras with different runs/game, well of course I thought weighted measurements take this into account and weight these events differently. But I already mentioned that previously and I guess you dont agree.

It seems that all you are asking for is a minor tweak for players who put the ball in play in high error environment. Not sure the current methods dont alrady do that. (Isnt there a co efficient for KOs in some of these measures?) And even so it has to be a very minor correction anyhow because there's very little difference in KO rates among batters in the dead ball era.
   107. Sunday silence Posted: August 27, 2018 at 08:39 PM (#5734522)
it's Ruth, Mays, Cobb, Bonds, Williams, Cy, and Johnson... with a few other names being "surprises" to the list (Wagner and Hornsby most notably on the offensive side and Clemens on the pitching side and of course Josh Gibson)


How in the hell does Cobb's name come up every single time in these discussions and Speaker never? That is truly baffling to me.

Cobb gets what about 15 hits more per year? And steals about what 25 bases more per year?
a
ANd SPeaker gets to what 75 fly balls more a year? (havent looked it up recently)

On raw numbers alone, you've got a problem there. So its odd that no one ever feels they can mention Speaker in these discussions. I mean I dont know if Detroit pitchers throw nothing but ground balls or there's a special wind current in Cleveland that carries all balls to CF...but I DO KNOW its bizarre that no one ever mentions Speaker's name.

You can do all the regression and use all these defensive metrics you want. the fact is current defensive metrics cant seem to agree more than +/- 10 runs on any single player, there's few players they agree on. You've got all these raw numbers telling you Speaker is some sort of prehistory Richie Ashburn or Undrew Jones and people seem to have collectively lost their minds on this...
   108. Sunday silence Posted: August 27, 2018 at 09:19 PM (#5734546)
EDIT: its probably like about 45 fly balls/year (would be more if 162 game schedule). its hard to figure they mostly dont have RF/9 inn. just per game.
   109. McCoy Posted: August 27, 2018 at 09:47 PM (#5734568)
this is the lynch pin of the entire reasoning. both yours and walts. and it makes no sense. consider these sentences;

(let's say its year 2000 to exclude barry bonds for a moment)

Ruth's HR rate was higher than anyone in history, therefore it would be extremely hard to believe he could improve upon that.

Does that make sense?

What about this one:

Ruth's strike out rate was higher than anyone in history (just assume that fact) therefore its very hard to believe it could get any worse.

You seem to accept that the second sentence cannot be true and yet you fight tooth and nail to deny the first sentence cannot be equally true by the same reasoning.

We could play this game a lot. "Bob Feller had the highest strike out rate in history, therefore its impossible to imagine him improving upon that.' yea or nay for that one?

Or how about Ty Cobb and stolen bases? What about Bill Mazeroski's fielding numbers? they can't go up in a different era?


I dont get it.


It would be a bad game to play. IF Ruth had the highest strikeout rate in history and 100 years later no one has come close it would be a pretty good assumption that it would be really hard to top those numbers. But everybody has topped those numbers. Thus there is ample proof that those numbers can be topped. Ruth hit more homers in year than everyone else combined. That hasn't happened again. Thus there is evidence that he would not do that again. Players do not routinely hit 100 homers nowadays. They do not routinely bat over .400.

Other than saying, dare to dream, there is no evidence that Ruth would become even more Ruthian in the modern game.
   110. Mefisto Posted: August 27, 2018 at 11:13 PM (#5734594)
If we're comparing say the 1920s lively ball era with modern times, isnt the run environment about the same? Somewhere around 4.5/ game? And if so then there would be nothing wrong with weighting doubles the same in both eras. Correct?


In fact, the run environment has stayed pretty constant in MLB for the past 100-odd years. What I'm pointing out is that the *reason* for that has changed quite dramatically. The linear weights formulas accurately predict how many runs will score, but they get used to say that a batter in 1905 "created" those runs, when in fact he didn't (and the batter in 2017 *did*). The context isn't being taken into account the way it should. If we want to compare batters from previous eras we have to account for the errors when we decide how much credit to give individual batters.
   111. Endless Trash Posted: August 28, 2018 at 03:36 AM (#5734625)
Is it any more ridiculous than the best composer ever writing 200+ years ago, or the best writer in the English language writing 400+ years ago?


Without question.

Across the board, athletes in all sports are faster, stronger, and better than the previous generation. This is easily demonstrated by the Olympics, where things are measured objectively and all the record holders are modern. Mike Trout is the best player today which means he's almost certainly the best player of all time. Its almost laughable to compare him to someone twenty years ago, let alone a hundred. And before people start, the reasons why don't matter to me at all. I know and don't care.

It is not comparable at all to the arts. Its a poor comparison.
   112. PreservedFish Posted: August 28, 2018 at 06:52 AM (#5734630)
The example of the Olympics is more fraught than you might expect. It’s very difficult to measure the effect of improved technology on those new records - even in simplistic events like the 100m dash or 50m freestyle.
   113. Sunday silence Posted: August 28, 2018 at 07:02 AM (#5734631)
IF Ruth had the highest strikeout rate in history and 100 years later no one has come close it would be a pretty good assumption that it would be really hard to top those numbers. But everybody has topped those numbers. Thus there is ample proof that those numbers can be topped.


You're missing the pt. of the analogy because you are focusing on the actual facts but I am making a larger point that is not directed at Ruth in particular:

OK now I have to look up stuff cause you cant understand this....

Ok it looks like Rob Deer has the highest KO percentage in history at least among those with decent careers, with 36%..Mark Reynolds currently active has 31% I dont know if someone will surpass Deer or not. THE POINT IS:

Lets assume Deer has the highest KO% in history, he did that playing in era that WAS NOT the highest KO era in history. IS THERE ANY REASON TO SUPPOSE THAT 36% IS THE ABSOLUTE LIMIT TO KO%?

or more general. Take any stat record set in an era where that era did not have the highest rate of whatever stat we are talking. Is there any reason to suppose that rate would be the highest possible?

OK do you understand this now? Ruth is playing in era which did not have the highest HR rate, but his HR rate is the highest. So why would we suppose that rate is the absolute limit to HR rate?

Sorry I didnt spell this out more clearly earlier.
   114. Sunday silence Posted: August 28, 2018 at 07:12 AM (#5734634)

The example of the Olympics is more fraught than you might expect. It’s very difficult to measure the effect of improved technology on those new records - even in simplistic events like the 100m dash or 50m freestyle.


Its a good point: if you compare Jesse Owens to say Carl Lewis, I think Lewis has maybe a step on him. But Owens is running in worse shoes on worse tracks its really hard to quantify it precisely.

Even more so in horse racing. Secretariat was really fast; but Citation running 25 years previously was also fast. But Citation is running on worse tracks, in shoes that were either steel or iron and not aluminum, and on tracks that were not banked or not as much. Its really hard to say if Sec. is any faster.

At 12 furloughs sure, Secretariat was unreal at that distance (see Belmont stakes) but say 9f or 10f.
   115. Sunday silence Posted: August 28, 2018 at 07:25 AM (#5734635)

Across the board, athletes in all sports are faster, stronger, and better than the previous generation.


This is simple and probably true and is not even what people are arguing. I dont think so. I mean obviously people have better nutrition and we assume training methods are better, and players are obviously bigger.

The ball is being pitched faster, we can argue what effect modern mounds, or balls might have, but I think in general they are throwing it harder. But I dont think anyone is arguing that.

BECAUSE IF THEY WERE THEN THEY'D BE CITING HEIGHTS AND WEIGHTS AND TIMES IN THE 40 OR TIME TO FIRST OR WHATEVER.

No one is doing that are they? They are talking about statistical records which are not measured on some absolute scale like seconds or pounds, but these are records set relative to other competitors. Someone is trying to throw a ball past you and you are trying to hit it, no one's arguing about the raw velocity of the ball.

the point of timelining or whatever they call it is to take some imaginary being with say Ruth's ability and transmit him into a different age.

This brings up numerous problems which Walt, for one, has alluded to. There are I guess several ways to assume things. I will suggest one way which is more favorable to my arguments:

Assume that we have an imaginary player who has the same ability RELATIVE to the LEAGUE as Ruth had, and then we plug him in and see what happens.

Well that's big assumption, cause no one today could possible dominate like Ruth. Well maybe not we cant be sure. But maybe Ruth was a freak too. Maybe he was a one in a billion freak.

I mean I dont think that's likely but if you're going to plug Ruth into the modern day, how else do you define his ability. Its relative to the league yes?

Well of course, no one thought of hitting HRs in 1919 cause it didnt seem to pay off, so for about 5 or maybe 10 years, you've got this HR explosion and you're trying to find players who can adjust to that. there werent school kids playing in 1918 who were developing a HR cut. those kids probably didnt start doing that until the mid 20s, so you've got this delay in the pipeline of HR hitters.

So then we assume that Ruth is playing in an era where everyone knows to swing for HRs, so he cant possibly hit more HRs than a team.

Well I dunno its all speculation in the end. It depends on what starting assumptions we agree upon.
   116. Sunday silence Posted: August 28, 2018 at 07:32 AM (#5734636)
If we want to compare batters from previous eras we have to account for the errors when we decide how much credit to give individual batters.


But you didnt answer my question, if the runs/game is the same in two different eras. Why wouldnt a double still be worth 0.75 weighted runs? I dont know why you are dodging this I thought I made that clear.


In fact, the run environment has stayed pretty constant in MLB for the past 100-odd years.


Im not even sure why you're making a point of this, because:

a) it really depends on what you mean by "pretty constant," and
b) its not constant it went up in the 20s, down in WW II, back up again, then down in the 60s, then back up again in 1973 etc.

This is like Gould saying the batting average in general has always been about .260 (which he did say in one of his books)

Its an idiotic point, it doesnt say anything, and if you look closely its not even true.

Sorry, Jay Gould really bugs the hell out of me for reasons...
   117. Rally Posted: August 28, 2018 at 07:51 AM (#5734638)
On Babe’s SLG vs Rob Deer’s K rate:

There is a difference between positive and negative records. There is nobody now who can match Babe’s slugging, nobody in history besides a Baroid Bonds who did it for a few years, not coming close on a career level. It’s reasonable to assume that barring some major change to the game nobody will approach his records in the future.

With Deer’s K rate, I know I could break it and so could you. There are plenty of minor league players who could do it. All it takes is a team willing to give that much playing time to someone who can’t make contact. The limit is on team tolerance not maximum human capability.
   118. McCoy Posted: August 28, 2018 at 08:35 AM (#5734651)
You're missing the pt. of the analogy because you are focusing on the actual facts but I am making a larger point that is not directed at Ruth in particular:

OK now I have to look up stuff cause you cant understand this....

Ok it looks like Rob Deer has the highest KO percentage in history at least among those with decent careers, with 36%..Mark Reynolds currently active has 31% I dont know if someone will surpass Deer or not. THE POINT IS:

Lets assume Deer has the highest KO% in history, he did that playing in era that WAS NOT the highest KO era in history. IS THERE ANY REASON TO SUPPOSE THAT 36% IS THE ABSOLUTE LIMIT TO KO%?

or more general. Take any stat record set in an era where that era did not have the highest rate of whatever stat we are talking. Is there any reason to suppose that rate would be the highest possible?

OK do you understand this now? Ruth is playing in era which did not have the highest HR rate, but his HR rate is the highest. So why would we suppose that rate is the absolute limit to HR rate?

Sorry I didnt spell this out more clearly earlier.


Yes, somebody could come along and have a 100% k rate or HR rate. Again, one can "dare to dream" but there really is no evidence that a guy from 100 years ago would actually play better now than he did 100 years ago.
   119. Rennie's Tenet Posted: August 28, 2018 at 09:23 AM (#5734671)
Is it any more ridiculous than the best composer ever writing 200+ years ago, or the best writer in the English language writing 400+ years ago?


I'm not sure they were the best. It seems obvious that they would just write all the best plays and symphonies first, leaving all the lesser ones for later. Hamlet obviously would write itself, and once its written it can't be written again.
   120. Mefisto Posted: August 28, 2018 at 09:43 AM (#5734684)
But you didnt answer my question, if the runs/game is the same in two different eras. Why wouldnt a double still be worth 0.75 weighted runs? I dont know why you are dodging this I thought I made that clear.


I answered it in 110. Let's take 2 actual seasons, NL 1911 and NL 1993. R/G in 1911 were 4.42. In 1993 they were 4.49. If we apply a linear weights formula to these seasons, ignoring my point about errors for the moment, we can use the same coefficients for the offensive events and we'll get the right answer for the number of runs scored in each season. It works for estimating league totals.

But people also use linear weights formulas for a different purpose. They use it to compare individual players. This is the point where the problem comes in. In 1911 the league ERA was 3.39. In 1993 it was 4.04. This means that while the run environment was the same both years, the runs scored in very different ways. In 1911 lots of the runs scored on errors, relatively speaking. In 1993, simplifying, lots of runs scored on HRs, relatively speaking (40/team in 1911, 140/team in 1993). In 1993 the batters had to do more work in order to score the same number of runs because the defenses had improved.

When it comes to comparing individual players (say, oWAR), the players in 1911 will be overrated because they're getting "credit" for the fact that defenders made more mistakes. The *reason* for the runs has changed even though the raw number has remained the same.
   121. Mefisto Posted: August 28, 2018 at 09:51 AM (#5734694)
Ok, found a great example. In 1911 Honus Wagner had 6.0 oWAR from a line of .334/23D/16T/9HR/67BB. In 1993 Jay Bell had a line of .310/32D/9T/9HR/77BB and yet had an oWAR of 4.8. Wagner didn't produce as many runs as he's getting credit for; those extra runs scored because the defense made more errors in 1911 than it did in 1993.
   122. McCoy Posted: August 28, 2018 at 10:02 AM (#5734702)
Wagner had a .930 OPS to Bell's .830 and you think Wagner should be getting more runs?
   123. McCoy Posted: August 28, 2018 at 10:44 AM (#5734743)
That should be "you don't think".
   124. Mefisto Posted: August 28, 2018 at 10:46 AM (#5734746)
I don't understand your question. (ETA: I posted this before I saw 123)

I think Wagner should be getting credit for fewer runs. I'm pointing out that the 2 players had substantially similar counting stats, but that Wagner got credit for more oWAR partly because using the same coefficients in both seasons "gave" Wagner "credit" for runs he didn't himself produce -- the defense just gave them to him (relatively speaking).
   125. McCoy Posted: August 28, 2018 at 10:48 AM (#5734749)
Wagner got 35 runs on offense to Bell's 20 because Wagner had 100 more points of OPS than Bell. Not because there were more errors.

   126. Endless Trash Posted: August 28, 2018 at 10:54 AM (#5734751)
the point of timelining or whatever they call it is to take some imaginary being with say Ruth's ability and transmit him into a different age.


I didn't quote your whole post because you're shouting like a crazy person. I haven't read the thread so I don't know what people are doing or not doing. I was responding to the person who responded to me, nothing more.

What you describe makes no sense. Everybody is a product of their environment. I don't even understand what you mean by transmit or ability. It doesn't make sense. If you took Babe Ruth and put him in today's game he would crap his pants. That's all I mean. If you're speculating about some other hypotheticals you're going to have to do a better job explaining it and not using all caps like a psycho.
   127. Mefisto Posted: August 28, 2018 at 11:27 AM (#5734766)
Wagner got 35 runs on offense to Bell's 20 because Wagner had 100 more points of OPS than Bell. Not because there were more errors.


That's not the way it works. OPS is a *rate* stat. oWAR is a *counting* stat.
   128. Rally Posted: August 28, 2018 at 01:40 PM (#5734919)
I think Wagner should be getting credit for fewer runs. I'm pointing out that the 2 players had substantially similar counting stats, but that Wagner got credit for more oWAR partly because using the same coefficients in both seasons "gave" Wagner "credit" for runs he didn't himself produce -- the defense just gave them to him (relatively speaking).


It's an accounting system. For it to balance somebody on the offense needs credit for creating those runs, even if they score on errors. Ideally credit goes to the player who reaches on error. If you don't have that, what I've done in the past is give some credit per non-strikeout out to account for the general error rate.
   129. Mefisto Posted: August 28, 2018 at 02:42 PM (#5734950)
As I see it, it's only an accounting system because the linear weights formulas are trying to account for total runs. They do that, and they do it very well. I just don't think it works to treat the same formula as an "account" when what we're trying to measure is the individual batter's credit for causing those runs in an error-neutral environment (that's what cross-era comparisons get used to do). Errors need to be taken into account or else we can't say that the batters "deserve" credit, so much as that the defense handed them the runs.

As an alternative, dWAR could be modified, though I think that's more complicated, and also we don't know enough about defense to make adjustments.
   130. Sunday silence Posted: August 28, 2018 at 03:26 PM (#5734964)
that's a really interesting example with Wagner and Bell. It looks like Bell is getting 20 runs in rBat; and Wagner 37 for nearly identical hitting numbers. baserunning is virtually the same w/ Bells 2 runs being negated by -2 on DPs.

Based on batting avg. it looks like Wagner may have had 15 or so more hits that might account for 6.7 runs. Other than that I guess its being on base while defenders are throwing the ball into the bleachers or whatever...

The run environment is pretty similar I think its 4.4/game in 1911 and 1993 cant be much different.

I have to think about it some more.
   131. Rally Posted: August 28, 2018 at 04:42 PM (#5735040)
Looks to me like Wagner has 17 more runs above average primarily because he made about 100 fewer outs.
   132. Zach Posted: August 28, 2018 at 05:48 PM (#5735087)
FYI ... Babe led the league in Ks 5 times; Thome 3. Babe K'd 12.5% in a 8.2% league (roughly 150 K+); Thome K'd 24.7% in a 16.9% league (a bit under 150).

I don't know about that. Babe was selling out for power in a league built around contact and speed. Thome played in a league where everybody was selling out for power, every pitch.
   133. Zach Posted: August 28, 2018 at 05:49 PM (#5735088)
I mean to say, I agree with the numbers, but I don't agree that Babe is a true talent 150 K+ hitter.
   134. McCoy Posted: August 28, 2018 at 05:54 PM (#5735090)

That's not the way it works. OPS is a *rate* stat. oWAR is a *counting* stat.


They are correlated. If you have 100 points of OPS on someone you are going to have more runs.
   135. Zach Posted: August 28, 2018 at 06:08 PM (#5735094)
There is a difference between positive and negative records. There is nobody now who can match Babe’s slugging, nobody in history besides a Baroid Bonds who did it for a few years, not coming close on a career level. It’s reasonable to assume that barring some major change to the game nobody will approach his records in the future.

Barry Bonds gives the lie to the argument that it's impossible to dominate a modern league the way Ruth dominated his era. It turns out it's extremely possible -- if you're an extremely good baseball player to start with, and extremely strong besides.

Barry Bonds 2001: .328/.515/.863/1.379
Barry Bonds 2002: .370/.582/.799/1.381

Babe Ruth 1927: .356/.486/.772/1.258
Babe Ruth 1928: .323/.463/.709/1.172

Height and weight were pretty comparable, too.
   136. Mefisto Posted: August 28, 2018 at 06:16 PM (#5735095)
They are correlated. If you have 100 points of OPS on someone you are going to have more runs.


That's an interesting thing. When Pete Palmer was working out some of this stuff, he first established the linear weights to predict total runs, and then noticed that OPS+ was correlated with that. But that demonstrates my point: OPS+ is correlated with a measure that over-credits an individual batter in an environment with more errors.
   137. PreservedFish Posted: August 28, 2018 at 07:22 PM (#5735120)
FYI ... Babe led the league in Ks 5 times; Thome 3. Babe K'd 12.5% in a 8.2% league (roughly 150 K+); Thome K'd 24.7% in a 16.9% league (a bit under 150).


I don't know about that. Babe was selling out for power in a league built around contact and speed. Thome played in a league where everybody was selling out for power, every pitch.


Right. If everyone in 1921 suddenly starts swinging Joey Gallo style, Ruth's K+ would plummet even though his Ks wouldn't budge in the slightest.
   138. McCoy Posted: August 28, 2018 at 10:06 PM (#5735209)
But that demonstrates my point: OPS+ is correlated with a measure that over-credits an individual batter in an environment with more errors.

Honus actually did more positive measurable things than Jay Bell did. Honus didn't have a higher OPS because of errors. In 1993 Fred McGriff had an almost 100 point OPS lead on Jay Bell and played in parks with a similar PF as Bell. He has 30 runs on offense to Bell's 20. Did McGriff somehow play in a pre-web glove environment in 1993 while Bell did not?


If you hit well you will produce more runs. This is boilerplate stuff.
   139. McCoy Posted: August 28, 2018 at 10:11 PM (#5735217)
Babe was not selling out for power. He took some mighty cuts in which he did sell out for power but the guy had a terrific batting average. If Ruth truly was selling out for power and he had the batting average and contact numbers he had that is really saying something about the pitchers of his era.
   140. Sunday silence Posted: August 29, 2018 at 07:17 AM (#5735321)
They are correlated. If you have 100 points of OPS on someone you are going to have more runs


Wait a second that's silly. You can be slugging 4.0 after one AB, that doesnt give you more runs that someone with say 10 AB or 100 AB. As he said, OPS is a rate, total runs or total runs created, that's a cumulative number.

Frankly that's elementary and I dont understand why you'd say this.
   141. Sunday silence Posted: August 29, 2018 at 07:33 AM (#5735322)
1993 Fred McGriff had an almost 100 point OPS lead on Jay Bell and played in parks with a similar PF as Bell. He has 30 runs on offense to Bell's 20.


OK you're on the right track but the specific facts here are also obscuring the main pt.

Wagner outslugged Bell by 70 pts in the aforementioned seasons, he also got on base 33 pts more. So that's at least part of the explanation for what's going on here.

But your example here you are using OPS which is a really bad way to compare hitters for two reasons:

1) its a rate stat and if you're talking about total runs produced in a season that's going to be totally weird unless both players had similar number of AB, which I guess they did, but also:

2) OPS is going to over state the value of HRs vs singles because slug counts a HR as 4 TB vs 1 for a single when in terms of runs created its: 1.45 vs .45. If two hitters are similar type hitters it probably doesnt matter, but when you compare a power hitter to a line drive hitter its going to over state.

Better to use weighted runs in a comparison like this, not sure how hard that is to find on baseballref site; Im having some issues there myself.

but like I said, its a good pt. that although Bell had similar number of extra base hits, Wagner is doing it with less outs. Wagner has somewhat higher .slug, ba and obp.
   142. Sunday silence Posted: August 29, 2018 at 07:45 AM (#5735323)
I don't know about that. Babe was selling out for power in a league built around contact and speed. Thome played in a league where everybody was selling out for power, every pitch.


That's a very interesting point, and one could test this idea also. Right? I mean as HR rates go up , more players would be likely swinging for the fences and hence KOs should go up. Right?

Except the connection between KOs and HRs is somewhat tenuous in Ruth's career.

(I am looking at totals for both leagues here, which I'd rather have AL but I wasnt able to pull it up easily) Looking at MLB rates:

HR rates pretty much doubled from 1919-1921, going from 0.2 per game to 0.4 (all numbers from here on are per game).

They stayed at that rate pretty much, there was a spike in 1925 (0.48) but they dont establish a higher rate until 1929 when it hits 0.55 and stays pretty much there up until WW II. (1942 brings a deadened ball owing to wartime restrictions on chemicals).

But KO rates dont correspond so much. They're about 2.8 in 1920 and pretty much stay that way until 1929 when they jump to 3.2. Thats about a 14% increase in KOs vs 37% increase in HRs. KOs again go up in 1936 to 3.4 and pretty much stay there until after WW II. It climbs steadily through the 50s and 60s.

So players in general dont seem to be changing their approach to hitting throughout the 20s. But something happened in 1929=30 and both HR and KOs do climb.

I know the NL and AL dont track exactly the same, so I'll try to find AL data. But interesting ideas.
   143. Sunday silence Posted: August 29, 2018 at 08:03 AM (#5735328)
Well continuing on this line of thinking produces some interesting facts. There is no way in hell you cant tell me the NL didnt deaden the ball in the 1931. HRs dropped by 45% and stayed that way till after the war. It wasnt until 1949 that the NL broke the league mark for HRs set in 1931. People keep saying the ball was never juiced or deadened but tell me how can you lose 45% of the HRs one year and it just stays that way for almost 2 decades? Bullsh!t.
   144. McCoy Posted: August 29, 2018 at 08:04 AM (#5735329)
This is kind of getting absurd. It's 2018 not 1997.

Fred McGriff and and Honus Wagner were better hitters than Jay Bell. We know this because they had higher OPS than Bell. We know OPS is a valid stat because, well, we've been doing this for decades and we know that OPS is a very good judge of hitting. We know Fred McGriff didn't have 10 PA and Bell 700.

If the numbers are wrong then what should they be? It is not enough to simply say they feel wrong.
   145. McCoy Posted: August 29, 2018 at 08:06 AM (#5735330)
People keep saying the ball was never juiced or deadened but tell me how can you lose 45% of the HRs one year and it just stays that way for almost 2 decades? Bullsh!t.

Well, whoever said that would be wrong. MLB back in the day would publicly tell you that they changed the ball. They tinkered with the ball publicly several times between 1910 and 1946.
   146. McCoy Posted: August 29, 2018 at 08:28 AM (#5735334)
COR throughout the years.

2017: .557
2016: .556 .494 Lichtman commissioned study in which the balls are fired at 120 mph-CCOR
2015: .558
2014: .554
2014-2015: .489 Lichtman commissioned study in which the balls are fired at 120 mph-CCOR
2013: .553
2012: .557
2011: .562
2010: .565
2009: .566
2008: .566
2007: .568
2006: .559
2005: .567
2004: .572
2003: .567
2000: .554
1999: .548
1998: .551
1982: .57
1980-1976: .53 to .56*****
1977: .563
1976: .556
1973: .559
1970: .556
1963: .559
1961: .5638 to .574 Popular Mechanics tested 12 balls as well and got a range of .50 to .68 while using a bat.
1960: .5517**
1953: .569
1952: .548**
1943: .42 for reclaimed rubber. .40 for balata*
1938: .46*
1936: .5672***
1927: .5534****
1925: .56
1923: .57
1914: .56


*: Briggs tested the balls by flinging them at 104 mph while MLB tests them at 58 mph
**: Balls tested were 1 year old.
***: 25 year old tested ball
****: 34 year old tested ball
*****: Exact year unknown since Charlie Finley's family donated them 25 years later.
   147. Sunday silence Posted: August 29, 2018 at 08:46 AM (#5735340)

Well, whoever said that would be wrong. MLB back in the day would publicly tell you that they changed the ball. They tinkered with the ball publicly several times between 1910 and 1946.


Last time I suggested deadening the ball and lowering the mound, Walt said deadening the ball has only happened once in the MLB so there's no proof it would work...SO thats where I got that.
   148. Sunday silence Posted: August 29, 2018 at 08:47 AM (#5735341)
COR throughout the years.


Can you please explain an acronym before introducing it?
   149. Sunday silence Posted: August 29, 2018 at 08:48 AM (#5735342)

If the numbers are wrong then what should they be? It is not enough to simply say they feel wrong.


I just said using weighted runs would be more accurate. Its RC+ or WOBA or something, it should be an easy look up.
   150. Mefisto Posted: August 29, 2018 at 08:51 AM (#5735346)
OPS+ works perfectly well most of the time. It works for players in the same season (obviously), and it works for players within a reasonable time frame. It's even somewhat useful in comparing players across longer time frames. What it doesn't do well is correlate with runs created over a long time frame because of the problem with errors.
   151. Rally Posted: August 29, 2018 at 08:51 AM (#5735347)
If Ruth truly was selling out for power and he had the batting average and contact numbers he had that is really saying something about the pitchers of his era.


Most of them probably threw 85 mph fastballs or less. Probably a small minority throwing 90+, and just a few elite who could throw 95. Part of that is smaller pitchers not having modern training methods. And part of it is a pitcher having to prepare to, and expect to throw 9 or more innings.

Imagine a hard throwing, Tampa Bay opener finding a time portal and going back 100 years. He would quickly be signed to a team, and blow away his opponents in the first inning. Then he sits on the bench, proud of his day of work as his hitters take their cuts. Then the manager tells him to get back out there. He's a bit surprised, but still able to pitch effectively for another inning or two. By the 5th inning he's gassed. Still throwing harder than his 1918 contemporaries, but off from his first inning velocity and the hitters have adjusted to him a bit. They take a few walks and put some balls in play, his fielders boot a few balls. By the 9th inning he's got nothing left. By now he's throwing the ball at typical 1918 velocity, just trying to get through the game.

3 days later his arm hurts more than it ever has before, but he's sent out to the mound again. He tries to throw hard and is able to for a bit, until he hears something pop. Damn, going to need that Tommy John surgery. But Dr. Frank Jobe is even going to be born for another 7 years. Too bad. His brief career is over. He's still bigger and stronger than most 1918 athletes. Maybe he can stick around as a hitter? He could hit a bit when he was in high school. Too bad the DH rule means he's only had 3 at bats since then.
   152. Sunday silence Posted: August 29, 2018 at 08:54 AM (#5735348)
1938: .46*
1936: .5672***
1927: .5534****


for which league? Clearly the AL and NL diverge quite bit after 1931.

In the 1020s, Its public knowledge at least I think it, that the NL didnt change their ball until after the AL did. It also seems clear that the NL ball got more lively in 1928 in contrast to the AL ball.

If these studies, if that's what they are, arent specifying which league I am not sure what we can make out of them. The fact that Pop Mechx cant replicate the results or whatever that's suggesting up there is also cpause for concern.
   153. McCoy Posted: August 29, 2018 at 09:02 AM (#5735354)
OPS+ works perfectly well most of the time. It works for players in the same season (obviously), and it works for players within a reasonable time frame. It's even somewhat useful in comparing players across longer time frames. What it doesn't do well is correlate with runs created over a long time frame because of the problem with errors.

Then show the work. Again, it is not enough to say it feels off.
   154. Rally Posted: August 29, 2018 at 09:03 AM (#5735356)
OPS+ works perfectly well most of the time. It works for players in the same season (obviously), and it works for players within a reasonable time frame. It's even somewhat useful in comparing players across longer time frames. What it doesn't do well is correlate with runs created over a long time frame because of the problem with errors.


I don't find this remotely convincing. Sticking with Bell vs. Wagner, OPS+ is looking at their hits, walks, extra base hits vs. outs. It compares them to the same for their leagues, and has nothing to do with errors. We know Jay Bell reached 4 times on error that season. Maybe Wagner reached on 20 errors, I don't know. Doesn't matter with that metric.

We know for certain that Wagner was better than his league by a greater margin than Bell was. We have good reason to assume than Bell's league had a higher level of play, but quantifying this is hard to do with any accuracy.
   155. McCoy Posted: August 29, 2018 at 09:07 AM (#5735360)
I just said using weighted runs would be more accurate. Its RC+ or WOBA or something, it should be an easy look up.

wOBA and RC+ are going to tell you the same story that OPS and OPS+ tells you that Fred and Honus were better hitters than Jay. They are all using the same data. You aren't going to get radically different differences because you switch to those. They'll weight things differently and present them in different forms and parameters

Fangraphs wOBA and wRC+
Jay Bell-.371 125
Fred MCGriff-.399 143
Honus Wagner-.439 155
   156. McCoy Posted: August 29, 2018 at 09:08 AM (#5735362)
I don't find this remotely convincing. Sticking with Bell vs. Wagner, OPS+ is looking at their hits, walks, extra base hits vs. outs. It compares them to the same for their leagues, and has nothing to do with errors. We know Jay Bell reached 4 times on error that season. Maybe Wagner reached on 20 errors, I don't know. Doesn't matter with that metric.

We know for certain that Wagner was better than his league by a greater margin than Bell was. We have good reason to assume than Bell's league had a higher level of play, but quantifying this is hard to do with any accuracy.


I believe he is saying that because of errors we can't trust Honus's run values and thus we cannot translate OPS (his offensive contributions) into a run value that is accurate. Not that OPS itself is flawed for Honus.

   157. Mefisto Posted: August 29, 2018 at 09:58 AM (#5735385)
156 is basically correct. Here's the historical sequence:

1. Pete Palmer develops linear weights measures of runs created.
2. Palmer notices that OPS/OPS+ correlates well with the linear weights results.
3. We all use OPS/OPS+ because it's quick and easy.

The problem is that if the linear weights formulas have the problem with errors that I've been pointing out, then the fact that OPS/OPS+ correlates with a problematic estimate of runs created is, well, a problem. Again, it's only a problem if we want to compare players over a long span of time. The declining rate of errors is pretty gradual, about .5%/year, so for reasonably short periods it's no big deal to use either OPS or WAR.
   158. Sunday silence Posted: August 29, 2018 at 10:19 AM (#5735396)
Mef: do you have an alternative value for RUns Created for Wagner 1911 and what would that number be? and how did you calculate it?
   159. Mefisto Posted: August 29, 2018 at 11:21 AM (#5735442)
I don't have a formal one, and I don't have the database to do it. What I have done for my own purposes -- I was trying to convince myself that I must be wrong about this -- is to reduce the estimate of runs created by the percentage of unearned runs for most of the top players in history. I didn't have the time or patience to do that year-by-year, so I used the average of unearned runs over their careers.

I just now looked at Wagner 1911, and I'd do it like this:

1. 23% of the runs scored in the NL were unearned.
2. Wagner's rBat shows that he is credited with 37 runs.
3. I multiply that by 77% to account for runs which actually scored on errors, which leaves him with an rBat of 28 runs. Note that I have to use the league average of unearned runs because there's no way I know of to calculate that for Wagner personally (or for any other player).

As I said, a lot of times the differences are going to be relatively small. For example, suppose we want to decide if Bonds or Mays was the better player. By BBREF WAR, Bonds has 162.8 WAR, Mays has 156.4. If we give Mays war time credit, he's probably around 168 WAR. But unearned runs accounted for 11% of total runs in 1954 and only 8% in 2004, so we should reduce Mays oRuns by about 3% relative to Bonds. At this point we throw up our hands and say this is all within measurement error anyway.

The big differences come with players from much earlier eras. In the NL from 1901-11, errors accounted for somewhere between 28% and 23% of all runs. That will require a huge adjustment for Wagner relative to a modern player like Mike Trout, because the rate of unearned runs in the AL last year was just over 9%.

To give you a sense for this, here's a partial list of players by adjusted batter wins using my estimate of actual runs created -- that is, runs which weren't the result of errors -- which I created in 2010. The second number is the adjusted batter wins shown today on BBREF:

Ruth 111.9, 130
Bonds 100.8, 122.5
Williams 90.5, 110
Aaron 83.8, 92.5
Musial 81, 94
Gehrig 79.6, 89.9
Mantle 79.3, 85.2
Mays 76.9, 84.2
Cobb 73.2, 110.1
Hornsby 72.1, 85.6

Wagner 46.7, 72.9
   160. McCoy Posted: August 29, 2018 at 11:54 AM (#5735457)
The Al and NL used the same ball since the teens I believed. They had a different name on them but they were both owned by the same company and the balls were coming out of the same factory.
   161. Rally Posted: August 29, 2018 at 11:55 AM (#5735458)
I've got more problems with this concept than I have time to post. But I'll start here:

%unearned runs is a horrible, awful way to estimate how many runs scored on errors. The further back in time you go, the worse it gets. The root of the problem is the definition of unearned runs.

Chucky Brown gets the first two outs in an inning. The next batter, Lucy, hits a line drive off the pitcher. It ricochets to shortstop Snoopy, but his throw to first is offline, and scored an error. Chucky walks the next two hitters and serves up a grand slam before finally getting out of the inning. Four runs score, all are unearned. But how much credit should really be debited from the batters? (pretending for a minute this is something worth doing) You really want to say the walkers contributed zero to run scoring that inning? Even the guy who hit the homer?

NL 1900, average team had:

1366 hits
(1079 singles
179 doubles
76 triples
32 homers)
379 walks
64 HBP

345 errors (total - a mix of reach on errors and advancement errors)

There were 5932 runs scored in the league, of which 4070 (68.6%) were earned.

Take the linear weight some of the positive batting elements. If you are going to reduce them by 31.4% because of errors, then how much weight does that imply you are giving to the errors? I get something like .83.

If they only gave out unearned runs for batters who reached on an error (in the Chucky Brown example above, 3 earned runs and one unearned) you might have something useful. But we don't.
   162. Slivers of Maranville descends into chaos (SdeB) Posted: August 29, 2018 at 12:07 PM (#5735463)

2. In #2, fielders make lots of errors. In this scenario, A will score X% of the time not because B did anything, nor because A did anything different, but due solely to something the defender did. Because A's double leads to more runs in an environment with more errors, A's double has a higher coefficient in a linear weights formula which purports to measure offense, even though the offense did nothing at all different than it did in scenario #1.


This assumes that errors are entirely the fault of the defender, does it not? But there's a difference with respect to the hitter to a short dribbler that even a gloveless fielder can get to easily and a hard-hit chopper that clangs off his glove. And if I know the third baseman is a poor fielder, and pull the ball so that it goes to him rather than to the slick-fielding second baseman, I ought to get credit for that, shouldn't I?
   163. Mefisto Posted: August 29, 2018 at 12:11 PM (#5735466)
@161: I think unearned runs is a definition as much as anything else. It's a "but for" test. In that sense, an error gives opportunities to the offense that it wouldn't have had otherwise. The overall run environment would be lower, in some cases much lower, without errors.

That said, I think it would be ok to use some other measure. My point is that I think that errors need to be taken into account somehow and at some level. However many runs one thinks scored "because of" errors, the coefficients should be reduced to reflect that. Otherwise we're saying that a single "creates" .45 runs, when that's true only for a given environment. Part of comparing players across eras is to make adjustments for the "environment", whether that's park factor or anything else.
   164. Mefisto Posted: August 29, 2018 at 12:16 PM (#5735469)
This assumes that errors are entirely the fault of the defender, does it not? But there's a difference with respect to the hitter to a short dribbler that even a gloveless fielder can get to easily and a hard-hit chopper that clangs off his glove. And if I know the third baseman is a poor fielder, and pull the ball so that it goes to him rather than to the slick-fielding second baseman, I ought to get credit for that, shouldn't I?


I think what you're saying is that harder hit balls are harder to field. The interesting thing is, though, that errors were at their peak in the Deadball Era and have declined continuously since batters began to hit the ball harder. As Chris Dial pointed out to me, a lot of that has to do with gloves. But I don't think that affects my larger point, which is that batters have to do more -- hit for more power, mostly -- to succeed today than they did in, say, 1905 because the defenses have improved so much (gloves, range, quality of field surface, positioning, etc.).
   165. McCoy Posted: August 29, 2018 at 12:26 PM (#5735477)
Rbat incorporates errors into its formula and if a player's exact numbers are not known they estimate. As I have said before linear weights incorporates errors into its tabulations. A back of the envelope cutting of runs is going to cause a heavy heavy altering of what the final number will be needlessly and erroneously.
   166. Mefisto Posted: August 29, 2018 at 01:21 PM (#5735533)
That's not really correct, and what BBREF does is close to the opposite of what I'm saying. Here's the BBREF explanation:

"For all seasons, we differentiate between strikeouts and other outs. In early baseball, pre-1920 or so, this is especially vital because error rates were high and DP rates were low, so there was a lot of benefit to putting the ball in play.

We include Reached on Errors for seasons that such data is available. For other years, we estimate the rate of ROEs and add that into our non-SO out values."

This is to say that BBREF is *crediting* the offense for runs scored on errors by "letting" the defense make errors (putting the ball in play, even if weakly, because the defense made so many mistakes). Modern players, in contrast, can't do that, so they have to hit for more power in order to generate the same number of runs.
   167. McCoy Posted: August 29, 2018 at 03:11 PM (#5735678)
You’re arguing that the hitter shouldn’t get credit for errors. Ok but having that view doesn’t mean rbat or any other linear weight is wrong or flawed. It means you have a different value or outlook on run scoring. I mean I could argue that batters shouldn’t get a full credit for walks or any at all for intentional walks. Having that belief doesn’t make linear weights wrong.
   168. Rally Posted: August 29, 2018 at 03:40 PM (#5735708)
OPS is going to over state the value of HRs vs singles because slug counts a HR as 4 TB vs 1 for a single when in terms of runs created its: 1.45 vs .45. If two hitters are similar type hitters it probably doesnt matter, but when you compare a power hitter to a line drive hitter its going to over state.


The above is applies to slugging percentage, not OPS. On base percentage understates the value of a home run since it gives that and a single a value of 1 each.

Add them together and you get OPS, 5 for a homer and 2 for a single. OPS doesn't have the perfect relative value of the events, but it's a lot closer than looking at only one of the two components.

   169. Rally Posted: August 29, 2018 at 03:44 PM (#5735714)
Jay Bell's 1993 season looks closer to Wagner's teammate, Chief Wilson. Wilson hit .300 with 34 doubles, 12 triples, and 12 homers for an .826 OPS. A little more power than Bell, a little less OBP. His Rbat was +17.
   170. McCoy Posted: August 29, 2018 at 03:52 PM (#5735726)
Chief's rBat comes out to 19.8 runs if given Jay's amount of PA.
   171. Ziggy's screen name Posted: August 29, 2018 at 04:23 PM (#5735754)
Is there a skill to reaching on errors? Given that a player reaches on errors in X% of Y PAs, can we with any confidence predict what percentage of future PAs will end in ROE?

This might not matter to the present discussion, but I'm interested anyway.
   172. Mefisto Posted: August 29, 2018 at 04:44 PM (#5735769)
@171: BBREF thinks there is (see 166). I'm skeptical, because I doubt that putting a ball into play weakly in the hopes that a defender will boot it, makes a productive AB. Putting the ball into play weakly is what Johnnie Lemaster did, not what Barry Bonds did.

If there were such a skill, it would obviously change in value over time as errors increased or decreased. Then, if we wanted to compare players across time, we'd need to account for that change in value.

If I were to put my view into a syllogism, it would look like this:

Errors are exogenous to existing runs created formulas.
Errors create runs.
Therefore, errors should be endogenous.

I want to emphasize yet again that this is solely in the case where we want to compare players from different eras. Existing formulas work perfectly well as long as the overall environment is substantially similar.
   173. PreservedFish Posted: August 29, 2018 at 04:45 PM (#5735770)
Is there a skill to reaching on errors?

There have to be correlations with speed, handedness, K%, GB%. Whether or not all that's a "skill" is an interesting semantic question, as nobody ever tries to ROE.
   174. McCoy Posted: August 29, 2018 at 04:51 PM (#5735776)
Ichiro consistently got on base because of errors whereas Adam Dunn did not. That was not a fluke.
   175. Mefisto Posted: August 29, 2018 at 05:08 PM (#5735791)
There was a lot of debate about ROE in some SABR Conventions in the mid 80s, specifically Oakland in 1985. There was a guy (forgot his name now; Davis something or something Davis) who urged it, even put out a book with a couple of years of data. He raised it in the panel with Bill James, Allan Roth and Pete Palmer. As I recall, there wasn't much enthusiasm for the idea. The main arguments against it were:

1. That wasn't how MLB defined OBP.
2. We lacked the data to figure it out for most years. [Note that Retrosheet has gone a long way to solving this problem and that BBREF uses an estimation to pick it up as I noted in 166]
3. It probably didn't matter much anyway. [Note the modern bias in this argument. It mattered a lot in 1905, but was considered to be insignificant in 1985]
   176. PreservedFish Posted: August 29, 2018 at 05:13 PM (#5735795)
Yeah, you'd expect Ichiro to be the perfect ROE guy. Then a fat high K righty to be the antithesis, Rob Deer types.
   177. Sweatpants Posted: August 29, 2018 at 05:23 PM (#5735803)
Yeah, you'd expect Ichiro to be the perfect ROE guy. Then a fat high K righty to be the antithesis, Rob Deer types.
I remember being surprised to learn that Harmon Killebrew had substantially more ROE than Tim Raines. If I recall correctly, right-handers have more ROE because they're more likely to hit it to short or third, and the throws from there are tougher.
   178. PreservedFish Posted: August 29, 2018 at 05:24 PM (#5735805)
Huh. Interesting.
   179. Greg Pope Posted: August 29, 2018 at 05:50 PM (#5735827)
If I recall correctly, right-handers have more ROE because they're more likely to hit it to short or third, and the throws from there are tougher.

I thought I remember that right handed power hitters have the most ROE. This is from something (The Book?) about 10-15 years ago. I don't remember the specifics. Yes, fast runners hurry the fielders more, but hitting the ball hard is more important.

Also, of course, errors are somewhat subjective. The SS may make an error because he's rushing due to a fast runner. But the scorer may see it as "he wouldn't have gotten him anyway" and award a hit.
   180. Mefisto Posted: August 29, 2018 at 06:23 PM (#5735841)
I doubt Rob Deer had many ROE, but I'm not sure it was because he was slow. Instead, he struck out a lot (reducing chances to reach base on error), he walked a fair amount (same), and he was a flyball hitter (ratio of GO/AO was only .56).

As a pure guess, I'd say that in today's game a guy with lots of ROE would probably hit into a lot of DPs (that being the other side of hitting hard ground balls). The contrast to Deer would be, say, Aaron. His GO/FO ratio was .84. He only struck out 10% of the time, and he famously grounded into a ton of DPs. This probably wouldn't have been true for a guy like Aaron in 1905 -- lots of those hard hit balls would have been booted and nobody turned very many DPs.

I suspect that if ROE were to be meaningful, it would have to be adjusted for (a) opportunities (that is, FB/GB ratio, Ks, and BB); and (b) overall context of errors and DPs (in order to compare across time).
   181. Zach Posted: August 29, 2018 at 07:15 PM (#5735885)
Mefisto: Why don't you just find linear weights coefficients that incorporate ROE at different points in baseball history?

You can do it in Excel, or with your favorite programming language. Just look at every team in, say, a 5 year window, and do a multivariable regression.

You're trying to find coefficients such that

A*1B + B*2B + C*3B + D*HR + E*ROE - F * outs - G*DP +H*SB - I*CS = Runs

I expect that you will find that E has about the same magnitude as A, since most errors result in the same outcome as a single. Minus a little because there's less chance for advancement, plus a little because players who are good at reaching on error tend to also be good at steals and triples.

If you don't want to do it yourself, I believe Tom Tango has linear weights values for almost every possible offensive event at different times in history somewhere on his site.
   182. Mefisto Posted: August 29, 2018 at 07:38 PM (#5735891)
A friend of mine suggested "Errors should in fact be endogenous to a model that purports to forecast runs created by a team, league or hitter (though I think I would argue that you should build such models separately for those three categories to avoid the reduction problem). That was my point of running of a longitudinal regression where the dependent variable is runs scored and the independent variables are all the different components, like singles, doubles, triples, home runs, walks and errors. That would provide a predictive equation across all eras. However, if the error terms were not IID (independent and identically distributed) that would suggest either a bias or more likely not a minimum variance estimator meaning some adjustment would be in order and/or be evidence that runs created differed in some fundamental way across the timeframes estimated. Technically, one should do this by estimating the model on part of the data and then using it to forecast results on another part of the data and determine the size of the error terms and any distributional characteristics that the error terms exhibit. Part of the problem in baseball stats is they aren’t really creating models, but rather they are trying to combine accounting terms that are basically tautological but that doesn’t really provide any of the dynamics of the underlying process."

I don't think I have the data to do this, or even your simpler suggestion. Is there a way (or place) to download it?
   183. Zach Posted: August 29, 2018 at 07:50 PM (#5735899)
The Lahman database would probably be your best bet.

http://www.seanlahman.com/baseball-archive/statistics/
   184. Mefisto Posted: August 29, 2018 at 08:13 PM (#5735918)
I can't believe I forgot about Lahman. Thanks. This may take a while unless someone wants to volunteer. :)
   185. Hank Gillette Posted: August 29, 2018 at 09:48 PM (#5736014)
Well that's big assumption, cause no one today could possible dominate like Ruth. Well maybe not we cant be sure. But maybe Ruth was a freak too. Maybe he was a one in a billion freak.


Guess it’s time to post this again.
   186. Mefisto Posted: August 29, 2018 at 11:12 PM (#5736061)
Jay Bell's 1993 season looks closer to Wagner's teammate, Chief Wilson. Wilson hit .300 with 34 doubles, 12 triples, and 12 homers for an .826 OPS. A little more power than Bell, a little less OBP. His Rbat was +17.


I'm not wedded to a particular comparison. I picked Bell/Wagner because the numbers were close. If they're closer with Wilson, that's fine.

The more important point is that BBREF rBat (see quote in 166) treats ROE as a *credit* for the hitters. It does that for Wilson and it does that for Bell. But here's the problem: the error environment in 1911 was highly inflated compared to 1993. It's similar to the way the run environment was inflated in, say, 1930. In order to compare players from different environments, we need to neutralize the errors just as we do for runs.
   187. McCoy Posted: August 29, 2018 at 11:32 PM (#5736073)
So Chief's production in 1911 should have resulted in him getting 11 rbat? That makes no sense.
   188. Endless Trash Posted: August 30, 2018 at 08:02 AM (#5736110)

I can't believe I forgot about Lahman. Thanks. This may take a while unless someone wants to volunteer. :)


I am more than happy to volunteer, but I am not 100% clear on what question we are trying to answer.
   189. Sunday silence Posted: August 30, 2018 at 08:28 AM (#5736123)
Of course reaching base on error is a skill; in the sense that some players are better at it than others. its not just a random odd occurance which the term "error" might make you think.

I think league average is 1%; Roberto Clemente had one of the highest rates at 1.9% or so; Willie Mays is about 1.3% etc. Just think about it; imagine a guy who's as slow as a slug an Albert Pujols or worse, and imagine a guy who's faster than Billy Hamilton, who's going to be involved in more close plays? Whose more likely to be safe if a fielder bobbles the ball? At a certain pt. you can be so slow that whatever the defense does is forgiveable they still get the out. so speed is important and i guess so is Right handed batting.

They call them "errors" which implies the fault entirely the fault of the defender but lets face it, if the runner makes the play close its more likely to be RoE (as well as a hit). It just goes to show that a certain pt. some plays are 50-50 they can be either errors or hits, it would likely be impossible to say conclusory is error or a hit. of course over time scoring is more forgiving so...



anyhow, Clemente seems to have been very quick on the base paths, i dont think he was flat out faster than Mays. Clemente did beat out a ground ball in inn. 8 of game 7 of the 1960 world series, I think just before that ball hit Kubek.
   190. Sunday silence Posted: August 30, 2018 at 08:32 AM (#5736125)

Mefisto: Why don't you just find linear weights coefficients that incorporate ROE at different points in baseball history?


This of course is a perfectly reasonable response to Mef's theory.
   191. Sunday silence Posted: August 30, 2018 at 08:38 AM (#5736129)
On base percentage understates the value of a home run since it gives that and a single a value of 1 each.

Add them together and you get OPS, 5 for a homer and 2 for a single.


I knew I messed that up right after I posted it. thanks for correction.

Its also gonna distort the value of BB's, right?
   192. Sunday silence Posted: August 30, 2018 at 08:42 AM (#5736130)
BBREF thinks there is (see 166). I'm skeptical, because I doubt that putting a ball into play weakly in the hopes that a defender will boot it, makes a productive AB


Jeezus! what kind of reasoning is this? its NOT PRODUCTIVE to deliberately hit a ball weakly, cause it's like 99% going to be an out. That doesnt mean that some player are 2% likely to reach base and other players are 0.5% likely.

the actual numbers are not important; whats important is your reasoning makes no sense here. on most ground balls the ba is what .220? call it .180 for "weak hit GB". Theres still gonna be fast players who get on by error or hit more than that and slower players getting on less. That's a function of batter footspeed (assume the ball is hit to the same place)
   193. Mefisto Posted: August 30, 2018 at 08:55 AM (#5736134)
I agree that footspeed is helpful if you hit a weak ground ball. I just doubt that hitting weak ground balls is a good thing.

That said, I have no problem incorporating ROE into WAR, *provided* that it neutralizes the error environment.
   194. Sunday silence Posted: August 30, 2018 at 09:04 AM (#5736138)
I expect that you will find that E has about the same magnitude as A, since most errors result in the same outcome as a single. Minus a little because there's less chance for advancement,


That's not true. Errors on GB are much more likely with men on base, Errors on FB are about 0.5% these are rather small compared to the other errors. Roughly (based on numbers that follow) errors on GB are 6x more likely than errors on FBs. That's not one I have a cite for, that's just back of envelope calc.

NOW, errors on GBs, in GENERAL, are 3.3% but that number dramatically goes up with men on base. I think it goes as high as 9% with 2nd and 3rd and less than 2 outs. Any time a double play is in order the rate pretty much doubles. I have a link to the general numbers I mention here but I've lost the one that shows the base/out situation, I can google it if you insist I know its out there.


There are several theories about why this is, one is that the runners in motion distract the fielders, another is that position the fielders in different places might cause this. I guess runners in motion are a distraction but also My own theory is simply that in the attempt to complete the double play, or anothe tough throw (say throwing home) the defender is rushed and makes an error (either fielding or throwing). If you run a cost benefit analysis of the value of a double play (in weighted runs) vs the downside of an error with men on base, it is still profitable to try for the double play even if say your error rate jumps to 9%.

NOTE FB errors dont rise or fall with men on base, they are more or less random events. That's also interesting, no one feels rushed trying to get to a FB with men on base.

So first mistaken assumption is to assume that GB errors are occuring at the same rate in all base/out situations. They are not. They rise much more with men on base. The result is quite important, I started to count errors with men on for last years playoffs, the average errors with men on base gives up about 1.0 weighted runs per occurance. Cause usually you have two runners moving up a base (approx 0 5 runs) plus getting to first 0.3 and not making an out 0.25.

So based on some back of envelope calc. roughly half the errors are made with men on base (worth about 1.0 or more wt'd runs) and the other half with no one on (about 0.4 or so).

the classic case is of course the pitcher or third basemen fielding a bunt and throwing the ball away with a couple or runners on base. This is basically like hitting the jack pot of weighted runs on errors. I think there were a couple plays like that in last years playoffs that were like 1.6 weighted runs each.

The upshot is very important. infielders making an error is probably about a 0.7 run differential on average, cause you have errors with no men on (about 0.4 runs) vs equal no. of errors with men on (about 1.0 runs). We'll do it again for this years play off as a check box event.

Here's another factoid that will challenge what you said. About 60% of errors are throwing errors (i read that somewhere I dont have a link) So like I said, throwing the ball away with men on base happens a lot more than you think if it was just a random event.

Which of course has all kinds of repercussions on defensive metrics, but since most of these are proprietary well...who the hell knows what they are doing?

I have made this point at least a couple of times in the past 6 months and usually it falls on deaf ears. As Ive said before my feeling is defensive metrics are under estimating the value of defense and this maybe one of the reasons.

thanks for READING if you got this far.
   195. Rennie's Tenet Posted: August 30, 2018 at 09:13 AM (#5736141)
I think just before that ball hit Kubek.


After that, just before Hal Smith's homer.
   196. Sunday silence Posted: August 30, 2018 at 09:20 AM (#5736147)
here's a good link to the ROE discussion:

https://www.retrosheet.org/Research/RuaneT/error_art.htm

In the above I dont know where he's getting that error rate number. I know Clemente reached on error like 180 times and he has him down for 140 for something..Oh: the "err" factor is how many MORE times they reached than one would expect normally. Weird.

Anyhow you can see its mostly fast guys, plus Joe Girardi. I dunno, was he fast?

Next here's the basic intro to the error discussion. NOTE this writer pretty much wiffs on the most important pt. he shows errors for base out situation but he fails to show the GB + base/out situation which is really dramatic.


https://www.beyondtheboxscore.com/2013/6/24/4456142/when-is-an-error-more-likely-to-occur-in-a-game

   197. Sunday silence Posted: August 30, 2018 at 09:21 AM (#5736148)

I agree that footspeed is helpful if you hit a weak ground ball. I just doubt that hitting weak ground balls is a good thing.


I dont think anyone is arguing that point. Are they?
   198. McCoy Posted: August 30, 2018 at 09:29 AM (#5736154)
Tangotiger when he looked at all errors did in fact put the run value of an error at roughly the same value as a single with the error being slightly more valuable.
   199. Mefisto Posted: August 30, 2018 at 09:46 AM (#5736166)
I dont think anyone is arguing that point. Are they?


Implicitly, yes. See the definition of rBat in 166.
   200. McCoy Posted: August 30, 2018 at 10:05 AM (#5736177)
Going to the other side
Page 2 of 3 pages  < 1 2 3 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

News

All News | Prime News

Old-School Newsstand


BBTF Partner

Dynasty League Baseball

Support BBTF

donate

Thanks to
phredbird
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogOT - Catch-All Pop Culture Extravaganza (November 2018)
(188 - 5:08pm, Nov 14)
Last: Man o' Schwar

NewsblogOT - November* 2018 College Football thread
(194 - 5:07pm, Nov 14)
Last: Howie Menckel

Gonfalon CubsNow what?
(224 - 5:05pm, Nov 14)
Last: Walt Davis

NewsblogSale of Baseball Prospectus
(50 - 5:04pm, Nov 14)
Last: QLE

NewsblogTime running out for officials to reach stadium deal with Rays
(2 - 5:00pm, Nov 14)
Last: Barry`s_Lazy_Boy

NewsblogZack Greinke trade makes sense for these teams
(9 - 4:56pm, Nov 14)
Last: Nasty Nate

NewsblogOT - NBA Thread (2018-19 season kickoff edition)
(2229 - 4:51pm, Nov 14)
Last: Mellow Mouse, Benevolent Space Tyrant

NewsblogMike Elias Orioles general manager | MLB.com
(15 - 4:50pm, Nov 14)
Last: Rally

NewsblogPrimer Dugout (and link of the day) 11-14-2018
(21 - 4:45pm, Nov 14)
Last: Ziggy's screen name

NewsblogDodgers pitcher Hyun-jin Ryu is only MLB player to accept $17.9M qualifying offer for 2019
(51 - 4:42pm, Nov 14)
Last: Baldrick

NewsblogProspects who stood out in Arizona Fall League
(1 - 4:31pm, Nov 14)
Last: Howie Menckel

NewsblogAtlanta/Southeast BBTF meet up
(67 - 4:19pm, Nov 14)
Last: stig-tossled, hornswoggled gef the typing mongoose

NewsblogOT: Wrestling Thread November 2014
(2312 - 4:14pm, Nov 14)
Last: rconn23

NewsblogOT: Soccer Thread (2018-19 season begins!)
(1197 - 3:35pm, Nov 14)
Last: Jose is an Absurd Force of Nature

Newsblog2018 Manager of the Year winners | MLB.com
(15 - 3:24pm, Nov 14)
Last: Khrushin it bro

Page rendered in 0.8103 seconds
46 querie(s) executed