User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
|
Demarini, Easton and TPX Baseball Bats
|
AllianceTickets.com has cheap MLB Tickets. Get all your Colorado Rockies Tickets, Seattle Mariners Tickets, San Francisco Giants Tickets and all your favorite baseball tickets here. We also carry cheap Denver Broncos Tickets, Seattle Seahawks Tickets and Denver Nuggets Tickets. |
For wholesale prices on baseball gifts and equipment, check these stores out! |
Page rendered in 0.9569 seconds
50 querie(s) executed
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.
I wouldn't either, my point being that Rivera, by some product of results and leverage, would have to be six times as good as Clemens to equal Clemens. (He only has to be three times as good as Bob Lemon, and perhaps that's within reach.) But if Rivera is the greatest closer ever and one can't come close to making that argument, that in itself says something about closing. It invalidates the "Mike Schmidt is a failed shortstop" objection, because it seems absurd to ask if Schmidt was as valuable as the weakest HOM shortstops.
No, that is not true at all. To equal Clemens, Rivera would have to be as good as Clemens. I have no idea what you are talking about with him having to be six times as good as Clemens.
If by leverage and performance he equals Clemens, he equals Clemens. There is no discount because he is a relief pitcher.
But if Rivera is the greatest closer ever and one can't come close to making that argument, that in itself says something about closing.
What does it say about closing? It says that no closer has been as good as Clemens. No C has been as good as Babe Ruth. Does that say something about the position of catching.
It invalidates the "Mike Schmidt is a failed shortstop" objection, because it seems absurd to ask if Schmidt was as valuable as the weakest HOM shortstops.
No, it doesn't. Labeling somebody a failed anything has no baring on any question.
Last year, Rivera has more win shares than Greg Maddux, does that mean you will not vote for Greg Maddux because he wasn't as good as a "failed starter."
All of that is irrelevant, and none of it has anything to do with "where in the list of the Trevor Hoffmans, Lee Smiths, John Francos, Rollie Fingerses etc. does it become clear that HOM consideration is no longer reasonable?"
Each of those persons have individual candidacies. They each had certain rates, seasons, comparisons, and total impacts of value. Each should be considered on those things.
But you don't start out and say:
(1) I'm going to ignore his rate stats.
(2) I'm going to ignore his unique achievements.
(3) I'm going to ignore how he compared with his peers.
(4) I'm going to make him have to perform with six times the value because I'm going to give him a demerit as "a failed starter".
(5) I'm going to give him every downgrade I can just because I don't like the fact that these guys actually created value.
I believe that Win Shares undervalues pitchers and even slightly more undervalues closers. Roughly, I think you have to give pitchers a 20-25% WS boost to find similar value hitters, and maybe 25-30
There is a ton wrong with WS. It is laughably crude. It as if James tried to come up with a magic metric and half way through said f!@# it, this is too hard I'm just going to start making stuff up so it will look right. I'm pleased with the batter offence part of WS. I think the reliever bonus is awkward and inutile. But I am embarrassed for James with respect to the defensive allocation of points (you know, the part where he randomly allocates percentage values for randomly selected accomplishments). Also, the 35/17/48 split (if I have the numbers right) is questionable to me. Generally, I think defense is overrated, since replacement level defence is not that far from average or above average defence.
No-one doubts that Rivera has less value then Clemens. But he is not competing with the top tier of HoM pitchers, he is competing with our borderline Rixeys and Fabers and Johns and Grimes' for a spot somewhere between the 50&#xil;e and bottom &#xil;e of our Hall.
Sure you do :) Rivera has pitched slightly more than 800 big-league innings. Clemens has pitched slightly less than 4,800. Rivera's got his work cut out for him.
Mike Schmidt had just over 10,000 plate appearances. Dave Hansen had about 2,000, and was a superb pinch-hitter, as pinch-hitters go. If they were equally good hitters, Hansen's opportunities would have to have been five times as important, or he'd have had to make proportionally more timely use of them, or something to make up the difference.
Now, Hansen was nowhere near as good a hitter as Schmidt, while going by ERA+ Rivera has been a significantly better pitcher than Clemens. So perhaps the analogy should be more like: Rivera is to Clemens as Dave Hansen in 1993 (batting .362/465/505 almost entirely as a pinch-hitter) was to Brett Butler in 1993 (batting .298/387/371 while playing every day) -- purely as hitters. Can the better rate performance in a few appearances match the lesser one in a lot more appearances? That's all that matters.
What does not matter is the special psychology of closing or the fact that there's a special name for that role on a pitching staff. There's an equally tough psychology of pinch-hitting, you could argue, and there's a name for the job. Nor, clearly, does it matter that closers get paid a lot more than pinch-hitters; the issue is not one of market forces.
How do you come up with these percentages?
Also, the 35/17/48 split (if I have the numbers right) is questionable to me.
Why? What would be more accurate splits?
Generally, I think defense is overrated, since replacement level defence is not that far from average or above average defence.
Why is defensive replacement not as far from average as offensive replacement?
These aren't rhetorical questions; I'm sincerely interested in understanding your thinking here.
I come to these conclusions by watching baseball for 30 years, reading James' books and countless other books and reading literally thousands of pages on this HoM part of the site over the last 3 years as well as reading the defense threads on the main part of the site. If we meet in Seattle next month I may get to discuss the reasons I have come to these conclusions at great length over a Girltini.
I sincerely look forward to it, and the Girltini is on me.
But WTF is a Girltini?
My drink of choice. It encompasses any brightly coloured martini that is typically ordered by females rather than males. I'm trying to popularize the word, so pass it on.
Mike Schmidt had just over 10,000 plate appearances. Dave Hansen had about 2,000, and was a superb pinch-hitter, as pinch-hitters go. If they were equally good hitters, Hansen's opportunities would have to have been five times as important, or he'd have had to make proportionally more timely use of them, or something to make up the difference.
Now, Hansen was nowhere near as good a hitter as Schmidt, while going by ERA+ Rivera has been a significantly better pitcher than Clemens. So perhaps the analogy should be more like: Rivera is to Clemens as Dave Hansen in 1993 (batting .362/465/505 almost entirely as a pinch-hitter) was to Brett Butler in 1993 (batting .298/387/371 while playing every day) -- purely as hitters. Can the better rate performance in a few appearances match the lesser one in a lot more appearances? That's all that matters.
I still have no idea what you are talking about.
(1) No one is arguing that Rivera is as good as Clemens.
(2) What matters is the value the player provided to his team in winning baseball games. Their constitution states that. That is not a rate of multiplier on opportunity. AND NO ONE HAS SUGGESTED THAT IS THE CASE. Rivera does not need to be "six times as good as Clemens." He needs to provide enough value to get in the HoM.
And even taking it down to a more suitable level. He doesn't need to be six times as good as Drysdale. He needs to provide enough value either in peak, career, or compared with his contemporaries, in whichever mix the electorate chooses to make him worth of enshrinement.
And to be honest, if someone were trying to look for a crude translation function for value, it would be assymptotic to account for bounds of human performance and opportunity.
What does not matter is the special psychology of closing or the fact that there's a special name for that role on a pitching staff. There's an equally tough psychology of pinch-hitting, you could argue, and there's a name for the job. Nor, clearly, does it matter that closers get paid a lot more than pinch-hitters; the issue is not one of market forces.
Huh?
(1) If a pinch hitter put up enough value to make the Hall of Anything other than the Hall of Pinch Hitters, then I'd support his candidacy. There hasn't been anyone on any measure that has put up anything close to that value.
(2) I've got no idea where you came up with most of that. It seems like you are introducing things that have not come up at all in this thread, not even from Treder's misdirection. There has been very little said about psychology at all. I'll discuss some of that supra. Thankfully with this crowd, they have not had to get into a discussion to recognize that the use of relievers has changed to maximize leverage, or to recognize that leverage exists at all. In many threads you would have to have hundreds of posts to even explain these simple concepts, or you would have someone who just wanted to argue that would throw FUD at them for just rhetorical enjoyment.
(3) A pinch hitter is going to see leverage of a maximum of about 5 for at most 1 at bat. They generally don't see anything near that high. A moderate closer, because of the role he performs, not because the team just drew a name out of the hat, is going to see an average LI of 1.9 for about 3.2 outs, and some of those will also be the 5s. Per appearance, a closer is already facing between three and five times as much leverage. On opportunity, your persistent closers will see about 70-80 innings, which is going to be about on e third of the highest starters. Considering that the average starter is getting about 150 innings, a closer is seeing opportunity at about 1/2 rate.
A pinch hitter, if they were only a pinch hitter and happened to pinch hit in every game, would by definition only be able to get 162 opportunities which is going to be about 1/5 of the league highest and about 1/4 to 1/3 of the average starter in opportunities. If they get more opportunity, it is going to drive down the leverage really hard and fast because those will be distributed in a way that leverage cannot be controlled.
There is no comparision between closers and pinch hitters. There is no comparison between closers and utility infielders.
And that psychology argument, I'm not sure what you think you discussing. The closest thing said by anybody is do they gross up closers numbers because the demand of the job require quick turnover. That demand may be both psychological and physical, but its a similar dynamic that also was once considered for catchers, middle infielders and center fielders. So its worthy for them to discuss. They can arrive wherever they choose.
But the psychological demand of closing is a personnel selection criteria, it is not a criteria that is likely to impact value.
And in a forum where no one has compared Rivera to Clemens, and then suggesting that Rivera needs to be six times as good as Clemens, does not seem to do anything but advance a bias rather than look for value.
It invalidates the "Mike Schmidt is a failed shortstop" objection, because it seems absurd to ask if Schmidt was as valuable as the weakest HOM shortstops.
No one thinks that is the case, either. What one thinks is that you shouldn't give Schmidt a performance minus because he couldn't make it at SS, and you shouldn't give a relief pitcher a performance minus because Steve Treder thinks that he is a failed starter.
Schmidt put up the value he did in a role that that team required. Rivera put up the value he did in a role the team required. Measure the value, and if there is anything about the role that causes you to make a consideration, then make it, but not because of some weird assertion that somebody is a "failed x"
How do you come up with these percentages?
A much better question is how did you come up with yours. You just read what James wrote. So, how did James come up with his. That's easy, he just pulled them out of his ass.
Why? What would be more accurate splits?
Why? Why are they accurate. This shouldn't be a place for rhetoric. There isn't "Bill James as a null hypothesis and then you argue against it." That is what I hope this place doesn't become. Its one of the things that Primer in general descended into because of certain people.
There question is "should we use it at all" And to be honest, it also features a few people that are more qualified as performance analysts than Bill James.
Why? What would be more accurate splits?
Why is that even a subject for a question of reliever value. If James is pulling numbers out of his ass, then there is no reason to even worry about what are better numbers, unless and until you have a thread where you are trying to come up with Win Shares V. 2.1456. Then the whole thing takes on a DIPSy quality. Why take something flawed and work on fixing it, when you can for any purpose you need, more easily create something that does work.
Why is defensive replacement not as far from average as offensive replacement?
These aren't rhetorical questions; I'm sincerely interested in understanding your thinking here.
Why is that a subject in Relief Pitcher valuation for the Hall of Merit. What you appear to be interested in is defending James.
Well ... OK. I'll be enjoying a singlemalt on the rocks, just the same.
Hey! That's my bit!
Does that mean "later"? Can't you just type "later"? It's the same number of keystrokes.
This isn't true, and I don't consider this a nit.
Every team has a third baseman *throughout history*.
the closer is a 20 year old creation. It isn't required, but a third baseman, for all intents and purposes, is.
There have always been thrid-basemen, but they didn't always do the same things, position themselves equally, etc...
And Rivera is a mensch! I'm simply not convinced that "closer" is an especially major role. It may be a popular role or a famous role, but it's a minor role.
The HOF was created by the media, for all intents and purposes.
Because they are the HOF voters, the media creates HOFers.
Media pressure is what caused managers to introduce the "closer role."
The "added pressure of the ninth inning" was also created by the media frenzy surrounding that "role."
The pressure is real now, though; performances of various relievers seem to leave no other choice. Also real, therefore, is that some pitchers can't handle it.
But the HOF "closer" is a media creation by premises 3 through 5 above, for enshrinement in a media creation by premises 1 and 2 above.
So the HOM, a creation of the SABR, should not follow the HOF's lead, it should follow SABR's. Therefore, pitchers are to be judged by their value as pitchers, which is limited by the number of innings they were able to pitch. Yes, that's limited by managerial usage of the pitcher, but bench position players, no matter how good, are subject to the same thing; we don't credit them for it, partly because of the sample-size issue. Well that issue exists for pitchers, too.
That said, the role (I don't want to say "position") has real value in the last 30 years and that value has been changing over that span.
As limited as may be the value inherent in the closer role, some men have managed to excel in that role - to the extent that they were much feared by the opposition and highly valued (and now highly remunerated) by their emplyers and teammates. Those few men, beginning with Hoyt Wilhelm, deserve their candidacy to be considered seriously - as seriously as any starting pitcher - and not judged as a failed shortstop or failed starting pitcher, or failed anything else.
BTW - I don;t think there is any evidence that a closer, or any other relief pitcher is a failed starter. At worst, a converted starter. Only the baseball gods, not us HOMer voters can really know if they failed.
By implication, are we failed members of the BBWAA?
I'm assuming you mean that we were created in the spirit of that group, but for others who are unaware, the HOM is not an offshoot of SABR (many of our voters are not members of that organization). We also have members who are of the "Orthodox Stathead" mode and not "acolytes" of James, Palmer, BP, etc. Of course, SABR itself is not monolithic in that respect, so maybe we should be annexed by them after all. :-)
But Johnson's late-blooming dramatic success is quite unusual among great starters.
One thing you learn at the HoM is how varied the paths of careers can really be - we have a lot of stories. We elected Dazzy Vance - left-handed, extremely late blooming (in effect, it seems he lost the front half of his career to arm injuries), and a dominant strikeout pitcher.
For another example that's been noted before around here. We elected both Amos Rusie and Joe "Iron Man" McGinnity. McGinnity was slightly older than Rusie; nontheless, McGinnity's career did not begin until Rusie's had ended (except for 22 innings of feeble belated comeback.)
He mistyped. He meant infra, which means "below"; supra means "above". Geez, Chris, how can you claim to be a chemist if you didn't pass Latin? You really need to stay current with the jargon here.
But Johnson's late-blooming dramatic success is quite unusual among great starters.
One thing you learn at the HoM is how varied the paths of careers can really be - we have a lot of stories. We elected Dazzy Vance - left-handed, extremely late blooming (in effect, it seems he lost the front half of his career to arm injuries), and a dominant strikeout pitcher.
For another example that's been noted before around here. We elected both Amos Rusie and Joe "Iron Man" McGinnity. McGinnity was slightly older than Rusie; nontheless, McGinnity's career did not begin until Rusie's had ended (except for 22 innings of feeble belated comeback.)
Are you attempting to make up for lost time, OCF? ;-)
For another example that's been noted before around here. We elected both Amos Rusie and Joe "Iron Man" McGinnity. McGinnity was slightly older than Rusie; nontheless, McGinnity's career did not begin until Rusie's had ended (except for 22 innings of feeble belated comeback.)
Red Ruffing is another hurler without the textbook HOF or HOM career.
There are quite a number of unusual paths to greatness, of course. Early Wynn also comes to mind.
But the fact remains that there is a textbook HOF or HOM career. Most great players establish themselves as standouts, if not as major stars, before they're 25, and this is certainly true of great starting pitchers.
Now as for relief pitchers: there have been a lot of guys who've had a terrific 3-year, or 5-year, or maybe even 7-year run as a reliever, and they were at all sorts of different ages when they did.
One oddity about the "closer" job: it's the only job on the team for which your replacement can "audition," live, on the major league team. You can't have a great year as a #2 SS or a #2 catcher - you're not playing enough to do that, unless the guy ahead of you is hurt. You can have a great year as a #2 starter, but that means you already have the job, and the team can keep both you and the "#1." But what happens when you show up, pitch in the set-up role, and pitch lights-out for a year or two? Consider the circumstances under which Wetteland left the Yankees and Percival left the Angels. The same sorts of decisions about established players (trade them or let them walk) are made all the time because of the potential of someone who's coming up behind, but that's potential. Rivera in 1996 or Francisco Rodriguez in 2002-2003 didn't represent "potential" but rather actual realized performance.
Actual realized performance in every sense, of course, except for the pressure (whatever it is, and it certainly exists to some degree) of protecting ninth-inning leads.
Very interesting observation. It undoubtedly is part of the explanation for the more rapid turnover among closers than among other roles.
And while it pains me to do so, I must admit that I really like the Ruby Tuesday's CoffeTini (or whatever they call it) and the Applebees ChocoTini (or whatever they call it). Both are excellent when fighting a hangover, for example.
I totally agree, Steve. There is a template that the majority of players do follow. But as OCF posted, the exceptions also matter.
For each decade 1960s thru 1990s, I found the 25 pitchers who started and 25 who finished the most games in the 10 yr time period. The ones who finished the most games I figure are the most-used bullpen aces. I could not use the 1950s this way, since there were some starters who also wound up with the most games finished; partly because of CG, partly because they did both, and lastly, because most teams did not have a bullpen ace. But from 1960 on, the lists were separate, even though there were some pitchers among the top ‘finishers’ who also started some.
The I found the ERAs for each pitcher, and averaged them (straight avg, without ‘weighting’ for IP). So, who would you think was better; the best starters, or the best closers? Here is the data:
Decade SP ERA RP ERA difference
1960s .. 3.26 ...... 3.20 ...... .06
1970s .. 3.29 ...... 3.10 ...... .19
1980s .. 3.69 ...... 3.21 ...... .48
1990s .. 3.81 ...... 3.32 ...... .49
If I eliminate the top 10 in each group, to throw out any potential outliers, and only use the 11th thru 25th best, the data becomes:
Decade SP ERA RP ERA difference
1960s .. 3.53 ...... 3.46 ...... .07
1970s .. 3.50 ...... 3.34 ...... .16
1980s .. 3.87 ...... 3.51 ...... .36
1990s .. 4.18 ...... 3.59 ...... .59
Pretty much the same pattern, but getting rid of Maddux, Clemens and the Big Unit hurts the 90s starters.
So, beginning around 1980, the relief aces began to be more effective (in terms of ERA) than the #1 starters in most rotations. Is it that managers aimed to cast some of their best as closers in the last generation, or has the lower IP/G usage made it easier on the relievers (see last bit of data belwo)? Or some other explanation?
Of the 25 1960s ‘finishers’ here, only Roy Face averaged less than 2 IP per G; most were 3 or over.
Of the 1990s finishers, many averaged les than 1.5 IP per G.
Well, the Bruce Sutter generation of relief pitchers were created because of the wear-and-tear that the earlier models had to put up with, so it makes sense that part of the increase in ERA+ for closers has to be due to the lighter workload. Beyond that? I haven't a clue, Tom.
As well as that better athletes tend to be selected for the role, most likely. Here's some of the data I presented in this article:
Average Top Save Producer
1960-1972: 94 IP, 1.68 IP/G, 130 ERA+
1973-1985: 97 IP, 1.64 IP/G, 140 ERA+
1986-1992: 77 IP, 1.30 IP/G, 146 ERA+
1993-2004: 67 IP, 1.08 IP/G, 149 ERA+
I have somewhere in my possession an article Bill James wrote on the subject in the early- to mid-80s; as a testament to my organizational skills, I have been looking for it for 15 minutes, to no avail (despite the fact that I refer to it rather regularly). It was in a SABR publication and re-published in a book that comprised several different SABR articles from a vast number of authors.
To the best of my recollection, James grouped starting and relief pitchers by various peripheral stats, such as K, BB, and H per IP, separating them into many groups (or "cells", as this was done on a spreadsheet or something). He then compared the starting pitchers in each group to the relief pitchers. The advantage RP enjoyed was rougly .25 of ERA. (As a totally hypothetical example, a starting pitcher who had 6 K/9 IP, 3 BB, and 9 H would have an ERA of 4.50, and a relief pitcher with the same rates would have a 4.25.)
One of Tango, MGL, and Andy Dolphin returned to the question in The Book. In this instance, they looked at how pitchers did who pitched in both starting and relief roles. For instance, Derek Lowe allowed a .260 wOBA against 896 batters as a starter and a .309 wOBA against 1133 batters as a reliever; they then weighted both numbers by the lowest PA between the two, added it up for every pitcher ... long story short, they found that the relievers' "advantage" is .80 of ERA.
It might be interesting to apply the two methods to the other study's population to see if they uncover comparable results.
***
The Book's study covers 1999-2002, but we might guess that its findings would apply about as well to pitchers from 1993 forward, given the near unanimity of the Closer Model by that point and the relatively similar offensive contexts assigned to the Juice Era. Mariano Rivera has a career ERA of 2.33 (entering this season) against a park-adjusted league average of 4.60, giving him an ERA+ of 197. Add .80 to his ERA (I'm considering his time as a starter negligible for this illustration), and his 3.13 ERA is an ERA+ of 147. (Of course, the league-average should probably changed if you were going to do this for everyone, shouldn't it?)
Career ERA+ marks for a few starters ... Greg Maddux is at 138, Randy Johnson 142, Roger Clemens 143. In simple terms of ERA+, Rivera may come out ahead of these guys, even if you account for the advantage he gains by being a reliever.
A few other relievers of the modern day: Trevor Hoffman's ERA+ drops from 146 to 113, Troy Percival's from 150 to 119, Billy Wagner's from 180 to 134.
Of course, we (well, I, in this post) haven't looked at leverage yet, and how to adjust a reliever's innings. Going back to Rivera, if we go ahead and hypothesize an LI of 1.9 (probably high), we can say that his 107 ERA+ in 806.7 IP "converts" to a 147 ERA+ in 1532.7 IP on the scale of a starter. This is the amount of innings a top starter might throw in seven years, so obviously Rivera does not compare to the clear HoMers of our time.
Tango had an interesting look at Sutter, Lee Smith, and Gossage through the eyes of LI, and concluded "The impact of an 80-inning reliever is no more than that of a 160-inning starter." But note that Tango did not adjust for the "performance advantage" we behold in RP; he simply multiplied all of the pitchers' numbers by the LI. (Perhaps it is a "double penalty" to do so?)
It may be better to compare to other relievers. Rivera's 197 ERA+ compares to the closer average of 149; his 2.33 ERA compares to an average of 3.09. By this reckoning, he has been 68 earned runs better than the average closer, coming into this season. Of course, runs saved by a closer are high-leverage, so if we apply a hypothetical 1.9 LI to Rivera's runs saved, we're up to nearly 130 (.161 ER saved against average per inning).
Just based on this thread, I've begun to do some research in a similar vein, that I'll be writing up as a THT article in a few weeks. I'm going to look at the full set of pitchers who've made at least 75 ML starts and 75 ML relief appearances over the past several decades. I've just begun putting it all together, but so far it's quite evident that not only do these pitchers as a group have better ERAs while relieving than while starting, they have distinctly better component stats as well (particularly hits allowed and strikeouts).
Just a few examples:
Phil Niekro 3.37 starter ERA, 2.97 reliever ERA
Joe Niekro 3.65, 3.06
David Wells 4.13, 3.23
Kenny Rogers 4.34, 3.26
Danny Darwin 4.04, 3.06
Tim Wakefield 4.35, 3.75
Mudcat Grant 3.80, 2.98
Derek Lowe 4.18, 2.98
Bruce Kison 3.79, 3.00
I wouldn't compare Rivera to other closers on performance - no more than I would compare starters to only #1 starters. I would compare Rivera's ERA to that of other relievers - all of them.
Actual realized performance in every sense, of course, except for the pressure (whatever it is, and it certainly exists to some degree) of protecting ninth-inning leads.
I'm sure that those around here who want to build a case for Rivera will be using 1996 as part of their case - as well they should.
I also remember seeing discussions of the Angels in the Percival-Rodriguez days to the effect that limiting Percival to a very strict closer role was putting Rodriguez into plenty of very high leverage situations, and that on a team level, it might make sense to put the best reliever on the team into the setup role, as long as the closer was still good.
It depends what you mean by great and by late start.
Earl Averill
Roberto Clemente.
The generalization may refer only to greater players than Averill and Clemente, the inner circle stars, in which neither Averill nor Clemente nor Vance, Ruffing, Niekro, or Leever qualifies.
OCF:
One oddity about the "closer" job: it's the only job on the team for which your replacement can "audition," live, on the major league team. You can't have a great year as a #2 SS or a #2 catcher - you're not playing enough to do that, unless the guy ahead of you is hurt. You can have a great year as a #2 starter, but that means you already have the job, and the team can keep both you and the "#1." But what happens when you show up, pitch in the set-up role, and pitch lights-out for a year or two? Consider the circumstances under which Wetteland left the Yankees and Percival left the Angels. The same sorts of decisions about established players (trade them or let them walk) are made all the time because of the potential of someone who's coming up behind, but that's potential. Rivera in 1996 or Francisco Rodriguez in 2002-2003 didn't represent "potential" but rather actual realized performance.
S.T.
Very interesting observation. It undoubtedly is part of the explanation for the more rapid turnover among closers than among other roles.
I agree: interesting, undoubtedly part of the explanation.
Some thoughts on its historical limitations:
Even now the setup role is not so well-defined as the closer role and that is not perfectly well-defined. Eg, many starting pitchers are starters more purely than Mariano Rivera, Troy Percival, or Robb Nen is a closer.
Historically, long relief and spot starts have been auditions for the regular starting pitcher role --for some long relievers and spot starters, the ones who are not yet failed starters. I think this must still be true to some extent, although development of the setup role as closer's sidekick limits the opportunities. (I suppose that "long relief" includes blowout relief whether long or short.)
For half the period the HOM project has covered, players auditioned for catcher by catching at the major league level, without benefit of injury to the team's best catcher.
Well, Joe, perhaps, but here's the thing. If indeed the sample of pitchers who both started and relieved quite a bit in the major leagues (at least 75 of each kind of appearance) is "biased towards pitchers who were better suited to relieving than to starting," then what sample wouldn't show such a bias? Obviously those pitchers who pitched almost entirely in relief would have to be considered as better suited to relieving than to starting, too. And those pitchers who pitched almost entirely as starters, while it certainly might be the case that they uniquely weren't better suited to relieving than to starting, their group also disproportionately includes all the pitchers most widely agreed to be the very best pitchers in baseball: that group would include Seaver and Gibson and Clemens and Maddux and all the rest. It just doesn't seem very plausible that these sorts of pitchers were only selected as starters because somehow their stuff wouldn't be effective in relief.
Moreover, the dataset I'm working with isn't small: it's 291 pitchers. I'm hoping that by starting with such a large group of pitchers, and limiting the smallest number of appearances as either a starter or a reliever to 75, then small sample size issues will be able to be minimized, even when I do the sorting along various axes, such as proportion of starts, strikeout vs. groundball, general quality, and so on.
Of course it won't be perfect. But if indeed a persistent significant trend is found to be common among a group of 291 pitchers over half a century, I think it will be hard to dismiss it as a nonrepresentative sample, but instead is reflecting something real.
"then what sample wouldn't show such a bias?"
Basically you are saying, "since we can't come up with an unbiased sample, I'm going to use this one, and even if the results are flawed, it's the best we can do."
I can't endorse that. If the only study that can be done is flawed, then we need to admit for the time being that the question can't be answered, at least not in this fashion. The process needs to stop there until something unbiased is thought of. It doesn't matter if it's 291 pitchers over 50 years or 10000 pitchers over a million years, if the population is biased towards being better suited for relieving, the numbers are going to show that - not that relieving is easier.
Again, I'm not saying relieving isn't easier, I just don't think this is the right way to go about studying it.
Plus, I don't see how it matters - at least not to the Hall of Merit, in terms of answering the question of who was more valuable.
I think for our purposes here, if we know how many runs an average or replacement level starter allows per inning, game, whatever . . . and we know how many runs an average or replacement level reliever allows, we simply compare the starters to the starters and the relievers to the relievers, and we know how much value each added. If relieving is easier, then the replacement level reliever is going to be a higher quality pitcher (in terms of runs allowed) than the replacement level starter and we can adjust based on that.
Actually, Joe, that's not at all what I'm saying. And I apologize for inadvertently giving you the impression that that was what I'm saying. But that isn't it at all, not even close.
If the only study that can be done is flawed, then we need to admit for the time being that the question can't be answered, at least not in this fashion. The process needs to stop there until something unbiased is thought of. It doesn't matter if it's 291 pitchers over 50 years or 10000 pitchers over a million years, if the population is biased towards being better suited for relieving, the numbers are going to show that - not that relieving is easier.
Joe, your a priori assumption is that the study is flawed. Yet you haven't established with any degree of sensibility whatsoever how the study is flawed. You've assumed it, but it seems that you haven't tested your assumption.
There is a finite actual population of real major league pitchers. It's far smaller than 10,000, and they haven't been playing for anything close to a million years. The actual population of major league pitchers working in a significant number of games over the past half-century is only a few hundred, and of that actual, finite population, 291 pitchers is a very meaningful, significant sample. It isn't a small, cherry-picked proportion, by any means. It's in fact a very large chunk of the population of pitchers who've ever pitched. The sample includes, in fact, 56,656 starts, and 55,757 relief appearances since 1957. To dismiss this huge population of pitchers and games without thinking it through isn't something anyone should do.
Again, I'm not saying relieving isn't easier, I just don't think this is the right way to go about studying it.
Well, Joe, then please explain specifically what's wrong with it. I repeat the question I asked you before: if this sample is biased in the direction of favoring relief pitching results, then what sample wouldn't be? Please tell me.
Because there are three groups of pitchers who've debuted since 1957 and who have pitched in at least 150 games:
- Those who have started at least 75 games, and relieved in at least 75 games. This is the population I'm examining.
- Those who have relieved in at least 150 games, but started fewer than 75. I'm not looking at them.
- Those who have started at least 150 games, but relieved in fewer than 75. I'm not looking at them.
Which of these three populations would not be biased in the manner you're asserting? How should I expand my selection to reduce/eliminate the bias?
Plus, I don't see how it matters - at least not to the Hall of Merit, in terms of answering the question of who was more valuable.
It may or may not matter to the Hall of Merit, although I can clearly see how it might matter very much. Why not take a look at the results of the study and find out? What about the study is so pointless? Please explain with more than "it's flawed."
Thank you!
- Those who have started at least 150 games, but relieved in fewer than 75. I'm not looking at them.
Okay, of course it's actually:
- Those who have pitched in at least 150 games, but started fewer than 75.
- Those who have pitched in at least 150 games, but relieved in fewer than 75.
And I'm not looking at either group.
TomH,
Did you break out the SP/RP splits for these guys? Are there any relief appearances among the SP's innings and any starts among the RP innings? Only asking because I think in the 1960s especially, many more relievers made starts than in subsequent decades, and that could depress the ERA of the relief group.
If the average closer is 0.8 runs better than a starter and the reliever replacement level is 0.3-0.5 runs better than a starter, I'd have to conclude that average closers are nearly a half run better than most relievers.
I would expect that when you break it out by starter-only and relief-only innings, the gap would widen a bit because starters' ERAs presumably get the reliever ERA boost when they relieve, and also that relievers, being more successful as relievers, would naturally see their ERAs go up when starting. In that era, there's plenty of Turk Farrell/Dave Giusti types who started for four years or less, but who were ace relievers for lenghty periods surrounding those starting years. Hoyt W is one, of course. We could call it the Rick Aguilera career path. I think Clay Carroll might have started for a year or two. But even more interesting is that many the 1960s relief guys tended to make a few starts here and there in most years until the late 1960s. Some like Fingers were just young pitchers finding their role, others like Farrell moved into the rotation then back to the pen.
That's definitely true and it would skew an analysis of relative ERAs if they are starting when their arms aren't stretched out.
Which is why it seems rather clear that the most valid comparison between starting and relieving performance is that of pitchers who performed both roles to a signficant degree. Most relevant of all will be the performance of pitchers working in both roles to a significant degree within the same year, or at least within the same portion of their careers. A guy who started during his peak but then pitched mop-up relief in his last few decling years isn't all that useful, nor is a guy who pitched long relief when breaking in before cracking the rotation.
Does that make sense?
I also want to say that finding the average ERA+ of all relievers as compared to starters would be valuable in order to recalibrate ERA+. So a 149 ERA+ would not actually be 149 in true value but it wouldnt' be average either, it would be about as valuable as a 130 or 125 or some because it would be compared to the ERA+'s of only other relievers. Then from this point we should levereage relief innings, giving us a pretty decent account of how much value a reliever gave his team compared to an average reliever AND a starting pitcher.
Does that make sense?
If not just ask...
Derek Lowe.
Phil Niekro, in 1967, threw 46 2/3 innings in 26 relief appearances (only five of which lasted three or more innings), allowing eight earned runs for an ERA of 1.54. He also started 20 times. In those 20 starts, he allowed just six total earned runs in the 60 innings that encompassed the first three innings of each start, for an ERA of 0.90. From the fourth inning of his starts on, as a starter, Niekro allowed 29 earned runs in 100 1/3 innings, an ERA of 2.60.
Of course, that's just one example - one pitcher, one season. My gut feeling is that if you look at just the first three innings of pitcher starts, and compare those results to the results from reliever outings, you'll likely see a pattern where reliever ERAs track the early ERAs of starters closely, with most of the difference between reliever ERAs and starter ERAs resulting from the bump to starter ERAs that occur later in the game.
-- MWE
I strongly suspect that's true. It's also the case that relievers rarely have to try and get the same guy out twice (or more).
Both of which are among the ways in which starting is a more challenging task than relieving.
...in the same game. Relievers often face batters more than once in a 3 game series. I think the truncation of games is where relievers win out. If someone hits a walkoff HR off you that's bad, but your'e not going to give up more runs. In other words, your liability is limited in a bad outing.
All true. But it's more than just an ERA-limiting thing going on here. The data I've looked at so far indicates signficant differences among pitcher's rate stats (H/IP, and particularly SO/IP) when starting vs. relieving.
David Smith, in a presentation at a SABR convention, noted that there appears to be a "batter learning" curve within a game, where a hitter is more likely to have a positive result the second or third time he faces a pitcher in a game. So I would expect that the rate stat differences are also likely to be the result of not having to go through the lineup multiple times.
-- MWE
Right, and the extent to which this is a function of the pitcher's increasing fatigue, and the extent to which it's a function of the batter "learning" the extent of the pitcher's reperetoire on that particular day, is impossible to disentangle, but the combination of the two is essentially the essence of what makes starting so difficult.
As well as, basically the basics, and fundamentally the fundamentals.
Unfortunately, I cannot recalculate that at the moment, but I did provide the year-by-year difference of reliever ERA to starter ERA in the bullpen optimization model. That still exists in some of the reliever utilization threads. If someone comes along that can google it, they will give you a link. Otherwise, I can re-run the queries tonight and show you. It won't take me a couple of months , and you won't have to wait for any forthcoming article to get it.
I'm not sure if the information will help you, because what you will find is that it is an increasing function. Relivers ERA compared to Starters ERA improves over time, oscillates and then improves again. The times of improvement are usually during the implementation in the innovation of reliver usage.
So if you tried to recalibrate in some way, you would have to reclibrate by era, and not universally.
You expect a closer to perform better relative to starters and the earlier Win share number bears this out, with a mean closer at 11 Win Shares, and a mean starter at 6 win shares.
I'm not sure why some people are conducting the aforementioned study. It really doesn't show anything that isn't known. Players should show better rate as a reliever than starter. The better you define their role, the further the difference.
That is already known by everyone, and has already been proven with other numbers.
The output of any such study would not increase knowledge in that area or prove any existing knowledge.
And more important, it would not lead to any help in developing a conversion mechanism. We already know that the specific use of the player can improve rate, and that is going to heavily depend on era. Knowing Zach Lowenlacy's starter and reliever era in 1951 doesn't help too much, until and unless you are voting on Zach Lowenlacy.
I expect it will just be the show of pretty numbers from which a which a non-supported piece of advocacy will be launched.
I thought you knew something about baseball history -- his name is Johnlee Lowenlacy.
I wouldn't be inclined to use Win Shares to demonstrate this, because of the way that saves and W/L are used to allocate WS to the advantage of the closer. The median closer, in 2005, had around 35 saves - the equivalent of about 12 wins for a starter (saves count as 1/3 of wins), in addition to his actual wins (which typically runs around 2-3 wins for a closer) - blown saves are not counted, probably to prevent double-counting with losses (although a good percentage of them don't turn into losses for the reliever who blows the save). The median starter, in 2005, had around 10 wins.
-- MWE
But first, something you'd expect. Here's 36 of the most prominent relievers of the 1960s era. They were chosen based on their appearnces, games finished and saves totals, as well as by being in my consciousness. In other words, I did my best to caputer the top 30ish guys in relief in that era.
What innings fall within the retrosheet-splits era are split into starting and relieving. As you can see, these guys threw almost 3 times as many innings in relief as in they did in rotation. As shown at bottom, their combined ERA is .85 points lower in relief than in rotation, very close to what we observed about all contempoprary relievers in TomH's chart above.
RELIEF PITCHERS OF THE 1960S ERAas a starter as a reliever
ip er ip er
--------------------------------------
wilhelm 383.33 136 1264 386
mcdaniel 394.33 221 1610.33 643
giusti 862 380 854.67 306
kjnowles 39.67 23 1052.33 356
hoerner 562.67 187
miller b 554.67 263 996.67 318
miller s 292.33 109 1014.67 309
face 8 3 986.67 337
fisher e 352.67 172 1186.33 470
wood 2119.33 921 565 209
abernathy 71.33 55 925.33 293
wyatt 44 30 643.33 260
bolin 1004.33 414 571.67 182
baldschun 704 290
granger 638.67 223
carroll 143.67 64 1209.67 378
farrell 928.33 353 772.33 294
perranoski 4 3 1170.67 362
sherry 97.33 48 702 277
kline 843 393 755.67 263
henry 3.33 1 591.67 195
radatz 693.67 242
regan 652 322 721 264
mcmahon 13.67 8 1297 474
upshaw 563 196
lee b 34.33 12 457.67 158
taylor 103 57 697 313
mcbean 494.67 212 577.67 218
linzy 5 4 812.33 254
aker 746.67 272
worthington 146 80 815 322
raymond 50 16 671 279
hall 339.33 147 763.67 243
locker 879 269
brewer j 179.67 87 860.67 268
gladding 5 0 595.67 209
======================================
TOTAL 10168.32 4534 29929.34 10519
<b>ERA AS SP 4.01 ERA AS RP 3.16</b>
OK, now here comes the weird thing.... I chose up 36 starting pitchers of the 1960s era to mathc the 36 relievers above. This was somewhat difficult because (a) quality SP tend to have longer careers than quality RP, so it's harder to find careers centered on this era (b) there are fewer quality SP by rate stats. Steve Barber and guys like that are all very much average-rate, mid-to-long career guys, but it was the best I could do once I got past the second tier. So I'm very aware that the quality of these two samples may not be identical; again, did my best.
When all was said and done, the ERAs for these 36 starting pitchers as starters and relievers were almost equal! What's more, they were ever so slightly higher in relief. This may be because many starting pitchers hurled relief either before or after their prime starting days ended. But by the same token, most starters in this era appear to make a handful of relief appearances a year (one or two usually).
as a sp as a rpinn er inn er
--------------------------------------
drysdale 3260.67 1066 72.33 29
gibson 3796.33 1376 88 42
koufax 2119.67 630 104.33 37
osteen 3363.33 1231 97.33 38
kaat 4137.67 1579 392.33 159
tiant 3348 1239 138.33 41
jackson l 2860.67 1062 139.33 46
ramos 1616.33 739 457.33 183
marichal 3476.33 1320 30.67 8
stottlemeyer 2654.67 874 6.67 4
short 2014.33 777 310.67 109
mcclain 1848.33 696 37.67 15
lonborg 2358.67 1015 105.67 41
lolich 3504 1339 134.33 51
lary 1548 630 81.67 23
purkey 1761.67 727 150 49
ellsworth 2020.67 847 134.67 43
mccormick 2129.67 868 244 111
sanford 1877.67 779 159 60
perry j 2956 1138 329.33 121
peters 1952.33 775 128.67 72
chance 1949.33 636 198 61
hands 1715.33 663 235.67 65
law 1765 678 108 36
veale 1764.33 585 161.67 71
bunning 3564.67 1289 91 18
podres 1720 684 119 45
mcdowell 2337.67 944 154.67 55
grant 1938.67 818 503.67 167
barber 1695.33 612 303.67 134
downing 2089.33 747 180 64
ford 2092 641 69.33 21
horlen 1867.67 646 133.33 45
jay 1323.33 544 176 81
johnson k 1522.67 565 215 103
maloney 1771.33 626 77.67 30
monboquette 1772 737 189.33 66
pappas 3082.33 1169 103.67 34
pascual 2414 894 79.67 22
peterson 2174 801 44.33 12
pizarro 1630.67 604 403.67 172
sadecki 2077.33 1002 423 204
siebert 1970.33 705 181.33 62
singer 2151.67 806 22.33 13
terry r 1667.33 652 168.67 78
wilson e 1991.67 819 60 23
======================================
TOTAL 104653 39574 7745.01 2964
<b>ERA AS SP 3.40 ERA AS RP 3.44</b>
Guys who were SPs in this era had an ERA of 3.40, guys who were mostly RP had a 3.16 ERA. So.24 points of difference---a little wider gap than TomH's data above, but not wholly inconsistent. In addition, if you take the starter and reliever innings and ER of the two groups and combine them, the 72 pitchers come up with a 3.46 ERA as starters and a 3.22 ERA as relivers. It's worth noting that the starters own 10 times as many starter innings as the relievers, but the relievers own four times as many relief innings as the relievers.
Okay, so my question now is this: should relievers be measured not against starters as starters, but rather starters as relievers? (I guess it's the same question if the ERAs are that similar for SPs in starting and relief...).
------------------------------------------------------
I hate to risk another backlasher backlashing, but I think that if one squinted, one might find some evidence of the failed starter theory in the ERAs of the starters being virtually identical in either role. Alternately, I hate to risk another Steve Treder post on failed starters, but backlasher could also find support for his theory of optimal utilization of a player's talents in the diminished effectiveness of relievers when they start. Same coin, different sides.
-- MWE
Theoretically, yes. The problem is, practically speaking, how do you do it? The problem with your sample of starters above is that very few of them pitched more than a handful of relief innings during their peak years: the bulk of their relief innings tended to come either when they were just breaking into the majors, and pitching long relief, and/or after they had declined, and were playing out the string in the bullpen. So your sample of relief innings by starters when they were capable of pitching at their best is small, as well as buried.
Which is why I maintain that the best test of the difference in difficulty between starting and relieving is to look only at pitchers who did a signficant amount of both, and within that group, paying particular attention to those pitchers who did a significant amount of both at the same portion(s) of their careers.
David Smith...noted that there appears to be a "batter learning" curve within a game, where a hitter is more likely to have a positive result the second or third time he faces a pitcher in a game. So I would expect that the rate stat differences are also likely to be the result of not having to go through the lineup multiple times.
I heard of this study, but couldn't the 'learning curve' be the 'pitcher is getting tired' curve? Maybe the ##s are the same, or maybe it will vary as managers with quick hooks get SPs out before they tire.
I'm not sure why some people are conducting the aforementioned study. It really doesn't show anything that isn't known. Players should show better rate as a reliever than starter. The better you define their role, the further the difference.
That is already known by everyone, and has already been proven with other numbers.
The output of any such study would not increase knowledge in that area or prove any existing knowledge.
It's premature to say we 'know' all of the exact numbers. But most of us are always learning. The study might not show a great deal, but it seems foolish to dismiss it a priori.
I can re-run the queries tonight and show you. It won't take me a couple of months , and you won't have to wait for any forthcoming article to get it.
Yes, BL, I'd like to see the queires. Without the accompanying annoying sarcasm, please.
-- MWE
It not only makes since, it presents the likely current decision fulcrum. Given this pitcher's skill sets, can they:
(1) Perform as a league average #2 starter or better.
(2) Can they perform as a league average closer or better.
If option 1 is strongly supported, then the pitcher is likely going to be made a starter, regardless of how you view him as a reliver.
If option 1 is weakly supported and option 2 is strongly supported, then you make him a closer.
If neither option is strongly supported then you look for another role. Then you see if they can perform the role because the first may have unseen injury consequences and the latter may have unseen psychological consequences. Albeit both of these are factored in the initial decision.
The biggest input on this decision looks to have been:
(a) Number of effective pitches. I can only think of few persons performing as a closer that had more than 2 effective pitches.
(b) Propensity for injury. Those that have weakened parts may be asked to not perform as much repitition.
Same coin, different sides.
The problem is that coin doesn't fit in your HoM vending machine. It doesn't matter if someone is a "failed starter", that is the issue.
And you have me wrong. It should be clear and without any rational rebuttal that:
(1) Rational managers seek to maximize a player's contribution; and
(2) Manager's ability to do that with pitchers has improved with time.
There isn't any measure that would contradict those assertions. Someone might try to muddy those conclusions.
The "other side of the coin" is that the player probably did not provide value doing other things, which includes: (1) Playing SS; (2) Hitting dingers; (3) Starting baseball games ...
You can play that negation with anything. You can say I'm a failed presidential candidate, because its a goal that many people (but not all) would like, and I have not yet been elected president.
The "edge of the coin" are players that have been successes at multiple things, e.g. Smoltz and Eckersly in starting and closing or Babe Ruth in pitching and hitting taters.
Its not that complex, as an aggregate class:
(1) Relievers outperform starters in rate stats.
(2) Starters outperform relievers in counting stats.
(3) The best relievers provide as much value as upper tier starters even with the change in rate.
On that note, I'll revise one of the earlier calculations. There was an error b/c I took the best RP rather than strictly closer, so I missed the ineptitudes of Looper and Mesa. I get the "closer" as previously defined with 9.7 Win Shares, which is still more than the average starter. The median reliver is still at around 2 win shares and the median starter is still at around 5 win shares.
If you define #1 starters as the one with the most win shares and the #2 as second most win shares, then the average for these are 14.7 for the #1 starter and and the number #2 at about 10.8 WS.
That was just the back of the envelope from 2005.
Lesson #247 on how WS undervalues pitchers. What was the average WS total for the top 30 position players in baseball that year? The top 60? Much higher than 15, for sure.
The WS book only goes up to 2000, so I only have Rivera's first 5 years on a year by year basis. I think all would agree that those five years are representative of his next five years as well. He finished first (or tied for first) among pitchers on the NYY 3 times and was third twice. Overall, he earned the most pitching WS for the Yankees over that five year period and over the ten year period of 1996-2005. Still, he only earns about 16 WS per year. That's Chad Curtis and Paul O'Neill territory.
I agree that the thought process very likely works in some similar way, but I'm not sure the choice is quite so defined. Managers do have different preferences -- some might make the change if the guy can't make it as a #4, others if he can't make it as a #2.
However, I'm not sure it matters for purposes of this discussion. In all but very rare cases, only #1 starters -- and only the best of these -- are likely to make the HOM. If the decision fulcrum is anything less than this, whatever value relievers have is unlikely to be sufficient for the HOM.
As I see it, the word "failed" has two meanings in this context. One refers to, say, me: I can't get anybody out even at my best. The other refers less to peak ability than to endurance -- he can retire batters, just not for very long. Starters have to do both, relievers only the first. "Failing" as a starter doesn't make one "failed" as a pitcher. I think Steve's basic point is correct -- it's implicit in your decision fulcrum point above -- but I'd probably use the word "unsuited" rather than "failed" in order to avoid the ambiguity.
If by "upper tier" you mean "#2 or #3 starters", I might agree. If you mean "#1 starters", I guess it's possible depending on the team. If you mean "HOM candidates", I'm dubious.
Cant' believe you placed Chad Curtis and Paul O'Neill in the same sentence there! ;-) Wasnt' O'Neill much better than that at his best and Curtis usually much worse? Also, I have no problem with most pitchers rating below postion plyaers in a stat like WS as in the modern era they don't pitch enough innings to have as much of an effect on a team as position players do. I mean pitching can't usually be more than 35-40% percent of the game anyway could it? So if more pitchers are splitting those innings, their WS totals will be lower. The question is how much should we adjust for this when if comes time to vote for them in the HOM. I think we have doen a decent job so far as neither Drysdale, Bunning, or Ford had numbers that I would have liked had they been OFers or SS's and all are in the HOM. And rightly so in my opinion.
BL,
I think the study would be useful not necessarily for the conclusions that would be made but in order to have actual numbers to help recalibrate ERA+'s.
And it is really annoying, childish, and even a little insulting to purposefully use different names for historical players and even posters, i.e. Jake Grasscock, Biz Markey, and Domino (Dimino I presume). You should be above that. Also, if you look at who we have elected I believe that SS is the most represented position followed by CF (I know we have a lot of OFers and I am not sure of the exact split), so could please stop calling this the Hall of Corner Players or whatever and do some homework?
I agree that WS isn't an especially good metric for relievers, but 16 win shares for Rivera may be about right. Win shares (and WARP) still misrepresent Rivera's value, however, by not having loss shares. Rivera would have had far fewer of _those_ than Curtis and O'Neill.
A stat like win shares does not adequately reflect the value of a very high rate of value accumulation, and this affects relief pitchers as a group more than any other group of players.
For those who have calculated LI, do they do the same for starters as well? I only ask because I would think that any starter that pitches the 7th, 8th, or 9th innings probably deserves some leverage since we are already giving leverage to relievers who pitch those innings. Does anyone know the answer to this question?
Their composite ERAs were 3.76 as starters, 3.40 as relievers. I made no adjustments for career path. That is I didn't partition them into early-career relievers, mid-career relievers, or late-career relievers. I'll do that next time out. Also, I made no adjustment for the presence of guys like Niekro, Niekro, Perry, Perry, Kaat, and others whose massive innings totals may be skewing the numbers in one direction or another. That also next time out. Anyhow, here's the guys I put together the data on.
aase
aguilera
alexander, d
bahnsen
barber
bird
bolin
candelaria
chance
cormier
darwindowning
drabowsky
drago
eckersley
escobar
farrell
fassero
flanagan
forsch k
fryman
giusti
gordon
gott
grant
hands
harris g
hawkins, l
hermanson
honeycutt
hough
jakcson g
jay
johnson k
kaat
kline
koosman
lamp
lockwood
lowe d
martinez d
may r
mcbean
mccormick
mcdowell s
merker
mesa
miller b
monboquette
morgan m
mulholland
murphy, t
niekro
niekro j
perry g
perry j
pizarro
power
purkey
ramos
reed r
regan
righetti
robinson d
rogers k
ruffin
russell j
sadecki
sanford
segui
short
siebert
smoltz
spillner
stanley
stewart d
swindell
terry r
tidrow
veale
wells d
wood
They just happenend to be two of the players Mo was adjacent to in the NYY WS standings in the years I was looking at. David Cone, David Wells and Andy Pettitte were other choices, but as wtih most starting pitchers, using a name out of context with a single year doesn't tell you much about the year they had. Whereas a late 90s Paul O'Neill should connote the feeling of a .285, 20 homerun, 100 rbi, 4th best position player on a great team type of season.
The dataset I'm working with is all pitchers with at least 75 starts and 75 relief appearances for whom we have Retrosheet split data (that is, with careers starting since 1957). It contains many of the guys above, of course. I haven't computed all 291 yet, but I've got most of them in there, and so far I've got a simple average of 3.99 starting ERA, 3.63 relieving ERA.
What I hope to look at isn't just ERA, but the component stats as well, and sorting them different ways, seeing if the effect gets stronger/weaker according to various independent variables, and indeed if the minority of pitchers who've been more effective as starters (and there are some) have something in common other than that. It promises to be fun, and who knows, we might even learn something.
The data I've looked at so far indicates signficant differences among pitcher's rate stats (H/IP, and particularly SO/IP) when starting vs. relieving.
David Smith, in a presentation at a SABR convention, noted that there appears to be a "batter learning" curve within a game, where a hitter is more likely to have a positive result the second or third time he faces a pitcher in a game. So I would expect that the rate stat differences are also likely to be the result of not having to go through the lineup multiple times.
- i think the "batter learning curve" is REAL obvious when you watch #4 or 5 starters. you might could say that lousy starting pitchers can't adjust to the batters adjusting to them
but about closer vs starter ERA - look at smoltz. it is easier to throw 10 - 20 pitches in 1 inning than 100 over 6-9 especially when you have to adjust in the longer games, and heat up and cool down and all that stuff.
would mo rivera even LAST 6 innings with throwing just a cutter?
and guys can throw harder if it is only for 1 inning. roy oswalt usually throws his FB 92 - 94 with an occasional 94. when he relieved, he threw 97. Which he did the last time he ever relieved in sept 2004. he threw like 10 FB - and nothing else for 3 Ks.
me i got this problem comparing the 2 different jobs - yeah they are both pitching a ball, but...
One of the interesting things I've seen so far is that even knuckleballers seem to have the relieving benefit (including P. Niekro, as you noted above), and at least from the few I've looked at so far, their K rates seem to be a lot higher in relief. That would suggest it isn't just a function of velocity.
That was what I was addressing in the previous post. ERA comparision of reliever to starter is an increasing function with oscillation in various periods. The study being proposed would not allow you to recalibrate ERA+.
The ratio varies with time, and time is not a component of the study.
And the problem that you run into is then you are way off from your mission. For instance, look at:
One of the interesting things I've seen so far is that even knuckleballers seem to have the relieving benefit (including P. Niekro, as you noted above), and at least from the few I've looked at so far, their K rates seem to be a lot higher in relief. That would suggest it isn't just a function of velocity.
That has no valuative component in it at all. Moreover, you aren't even seeing data, you are seeing date being teased.
I agree that the thought process very likely works in some similar way, but I'm not sure the choice is quite so defined. Managers do have different preferences -- some might make the change if the guy can't make it as a #4, others if he can't make it as a #2.
However, I'm not sure it matters for purposes of this discussion.
You are correct, it makes absolutly no difference for this discussion, but there is a force that keeps yanking the discussion into this direction.
And I agree, I'm sure the actual decision mechanics vary, and I would opine they display less variance as time increases.
By "fulcrum", I am not suggesting that every person makes a decision on this basis, I am suggesting this is the point of decision for utility maximization. And right now, its a hypothesis, its not even an assertion.
In all but very rare cases, only #1 starters -- and only the best of these -- are likely to make the HOM.
And that shouldn't matter either. #1 Starter, as is used here, is a distinction of selection, not value. If you use #1 Starter for value, its going to be very rare that any pitcher is the #1 Starter every year in his career. That distinction might have some merit in the discussion for what these guys term "peak" if and only if, the #1 starter isn't playing for creampuffs. There were a lot of #1 starters that aren't even going to sniff the Hall of the Played for a little while.
What is being selected is pitchers, based on the value they provided.
Also, if you look at who we have elected I believe that SS is the most represented position followed by CF (I know we have a lot of OFers and I am not sure of the exact split), so could please stop calling this the Hall of Corner Players or whatever and do some homework?
I have no desire to repeat the sillier conversations that take place on the main board. But there is no issue of homework. And the Hall of Corner Players is what you would end up with, not what you have, if you use the warped logic being offered that has you discriminate against a class of players. And you justified that be looking at how positions compare in WS. No pitcher appears at all in the top 25 for WS last year.
If you want to see value of closers to the staff, its very consistent with Mike Emeigh's assertion.
Last year, using WS, seven teams had closers that earned as many or more win shares than the #1 starter; a total of fourteen teams has closers that earned as many or more win shares than the #2 starter. Last year, the best couple of closers are performing at Halladay, Peavy and Big Eunich level.
And when these guys start stringing together years, that is a lot of value.
No, because that presupposes value that may not be valuable.
You can say the same thing about track athletes. Are all sprinters failed long distance runners? Are quarter horses failed animals?
Suitability is not only the correct word, it is the word that has more meaning.
Some people are good enough to win gun to gun, but many races are won in the bell lap.
The lowest in the Top 60 position players is 21 WS; you don't even have to go to mean.
Moreover, their is only five or six pitchers in the Top 60 at all.
If you look at each team individually (and giving .5 points for ties), there are only seven teams that have a non-corner WS leader (.5 C, .5SP, 1.5 CF, 2 SS, 2 2b).
So Taguchi has more WS than any pitcher on ten teams, and the same amount as likely future HoF Tommy Glavine. If not for a couple of closers named Wagner and Turnbow, it would even be higher.
and guys can throw harder if it is only for 1 inning. roy oswalt usually throws his FB 92 - 94 with an occasional 94. when he relieved, he threw 97. Which he did the last time he ever relieved in sept 2004. he threw like 10 FB - and nothing else for 3 Ks.
One of the interesting things I've seen so far is that even knuckleballers seem to have the relieving benefit (including P. Niekro, as you noted above), and at least from the few I've looked at so far, their K rates seem to be a lot higher in relief. That would suggest it isn't just a function of velocity.
- well then i think it is part - the hitters don't have time to adjust - and maybe part the pitcher does not have to pace himself.
i know a knuckleballer only throws one pitch, but with other guys, well maybe also they can get away with 1 good pitch (like trevor hoffman and his changeup)
Backlasher Posted: June 05, 2006 at 06:07 PM (#2052288)
I think Steve's basic point is correct -- it's implicit in your decision fulcrum point above -- but I'd probably use the word "unsuited" rather than "failed" in order to avoid the ambiguity.
No, because that presupposes value that may not be valuable.
You can say the same thing about track athletes. Are all sprinters failed long distance runners? Are quarter horses failed animals?
Suitability is not only the correct word, it is the word that has more meaning.
Some people are good enough to win gun to gun, but many races are won in the bell lap.
- actually i think that you comparing sprinter to LD runner is a good one.
you don't take the time for a sprinter and multiply it by 5 miles or something to get what times "should be" for a distance runner. you can't just say - why not, they both are running, right?
it makes more sense to compare SP to SP and closers to closers. maybe we shouldn't say - well, they both pitching right???
- anyhow, now i gotta go and tie down all the backyard stuff because we gonna have hurricane force winds here in 90 minutes caused by the combined sucking of my stros and the cubs...
OTOH the first inning is the highest scoring inning, presumably because that's when the best batters bat. I suppose the learning curve also could involve the park.
Your data in post 372 and similar posts is interesting, but I wonder if the analysis was done correctly. In averaging over all pitchers, it is not proper to split each pitcher into starting and relief IP and then to add up the totals of the two sets, because the relative weights of the different pitchers in each set changes.
For example, if pitcher A had a 3.00 ERA in both 120 innings as a starter and 60 innings in relief, and pitcher B had a 6.00 ERA in 60 innings as a starter and 120 innings in relief, your analysis would show the starters with a 4.00 ERA and the relievers with a 5.00 ERA, implying that the pitchers had worse ERAs in relief, even though this does not hold for any individual pitcher.
I suggest, though it involves more work, and probably would not qualitatively change your conclusions, that it would be better to take each individual pitcher's split and to prorate the part with the greater innings pitched down to the number of innings pitched of the other part before adding everything together.
I almost made this analogy; it's a good one. Yes, in one sense of the word, marathoners are failed sprinters. They, of course, would prefer that we say how well suited they are to marathoning.
As to value, that's harder to say. If you're chasing the giraffe, the marathoner has more value. If the lion is chasing you....
What was being discussed by some of the Merit boys was how much of this to do. In just about any case, they are comparing:
(1) Position to Position across time
(2) Position to Position from Era to Era
(3) Value to Value on all sets of players
(4) Rate to Rate
Some may not admit they were comparing position to position, and just say "they were accounting for the defense" but considering that none would be able to observe the defense and their ability to quantify that defense is suspect, they were de facto comparing position to position.
They have a few more tools, some crude, some more exacting when doing the pitcher comparsions, so that is why they have this thread. The problem that I mentioned is that one person wants to devaule all benefits and increase the value of all drawbacks when making this comparison.
you don't take the time for a sprinter and multiply it by 5 miles or something to get what times "should be" for a distance runner. you can't just say - why not, they both are running, right?
Yes, and the reverse of this is what is being offered in the "study" that is being teased and pimped. It would be the equivalent of taking Michael Johnsons' time in the marathon, and then devaluing his actual 200 race performance.
Then taking a marathoner, and assume that they would improve by the mean difference between the 200 split differential taken from anyone who ran a marathon and every single other race they ever ran.
There is no question and no issue that most people will run a 200 meter dash quicker than their mean split time for 200 meters in a marathon.
While it's certainly possible that pitchers are sharper, according to Smith (via Emeigh), the numbers don't show that on a hitter-by-hitter basis.
Yes, a simple average isn't really proper. Most direct and proper is just to compute the aggregate ERA of both sets based on the total innings and earned runs, which is what I plan to do.
You can say the same thing about track athletes. Are all sprinters failed long distance runners?
I almost made this analogy; it's a good one. Yes, in one sense of the word, marathoners are failed sprinters. They, of course, would prefer that we say how well suited they are to marathoning.
- well their muscles are different, aren't they?
As to value, that's harder to say. If you're chasing the giraffe, the marathoner has more value. If the lion is chasing you....
- grinning
well sure
as my granma always said, a place for everything and everything in its place.
or was it
sometimes u feel like nut, sometimes u don't
starting pitchers go a whole game
but the closers won't....
As to value, that's harder to say. If you're chasing the giraffe, the marathoner has more value. If the lion is chasing you....
And in this case, the sprinter is the closer and the marathoner is the starting pitcher if you make the analogy hold true.
And for the second analogy its also proper. In the first few innings you are chasing the giraffe of nine innings. In high LI situations, that lion is biting on your ass.
And without the giraffe you don't eat dinner, but if you don't get away from the lion, you are dinner.
Proper person at the proper time.
And for all the aforementioned reasons, it will not tell you anything at all on any value question.
It will not even give you a translation, because to use Field's analogy, you will be combining 10K runs with 100 m sprints and assuming they are all the same thing. And even if you divided them up, it will not tell you anything about the specific pitcher's skill to maximize one environment.
I would think it's a given that the first inning is the highest scoring because you are almost always facing the other team's best hitters?
what inning did the 1961 yankees score the most runs?
Away games, the Yankees scored their most runs in the 4th (59). The 1st inning was second-most with 53. Home games, the Yankees scored their most runs in the 1st (53 again -- interestingly, tied with the 2nd inning). Overall, the 106 runs they scored in the 1st was the most of any inning.
On the road, they scored 32 runs in the 2nd inning, lowest of any except the 5th (29).
Boyer batted first in 26 games, second in 11, third (!) in 3, seventh in 15, and eighth in 93.
Lopez batted second in 28 games, third in 8, and seventh in 24.
Kubek batted first in 18 games, second in 115, third in 2, seventh in 6, and eighth in 9.
It's a lot easier to say this in hindsight than it would have been at the time, but they probably just should have had Mantle lead off, Maris, second, Howard third, and Berra fourth, and drawn the rest of the names from a hat.
You must be Registered and Logged In to post comments.
<< Back to main