Sorry for the delay!

Interesting results, I'm kind of shocked that we rated Smokey Joe ahead of Paige.

One voter did leave Paige off, but he would have had to have ranked him 9th for him to have passed Williams.

How did I miss that someone didn't rank Paige? I was the consistency police for this election, sorry I missed that one.

There's no way Satchel Paige wasn't one of the top 20 pitchers of all time.

That was karlmagnus. He also had Spalding and Caruthers ahead of Seaver and Spahn (and Caruthers, Keefe and Radbourne on his ballot but Clarkson not), Nichols ahead of Grove, and so on. I didn't do any kind of consensus calculation with this, but karl was pretty clearly the outlier.

The voter who omitted Feller was JPWF 13 and the voter who omitted Hubbell was AJM.

With the striking gap between #18 and #19, it's a little surprising that only four voters had all of the first 18 on their ballots: Mike Webber, Al Peterson, Joe Dimino, and me. For us four, who were the other two? Mike Webber had Dihigo and Rogan; Al Peterson had Walsh and Keefe, Joe Dimino also had Dihigo and Rogan, and I had Blyleven and Plank.

One comment about the ranking of Negro League pitchers: this particular sample of the electorate was missing several of the people who did the most for our understanding of those leagues. Notably, neither Chris Cobb nor KJOK participated.

"I'm kind of shocked that we rated Smokey Joe ahead of Paige."

I'm not.

Satchel was incredible in his early years, had a dead-arm/barnstorming phase, then an amazing late-career even in MLB phase.

Smokey Joe not only pitched like Grover Cleveland Alexander, he pitched against him and beat him.

Seemed to me that the consensus had Smokey Joe in that company, without a mid-career dip that I know of.

the voter who omitted Hubbell was AJM.

He was real close to making it. In fact I think I may have overlooked him, but I don't think I would've put him over Paige anyway (who was #20 for me).

"I'm kind of shocked that we rated Smokey Joe ahead of Paige."

I'm not.

Same here for all of the reasons that you mentioned, Howie.

I passed this list along to a friend of mine. He's not exactly a sabermetrician but he does listen to me. I got back a bit of a rant about how could we put Blyleven ahead of Koufax, but I also got this:

I can imagine that exhaustion has set in -- but even so, this is badly
skewed to the early history of baseball.

We've been over this time and time again, and it's always difficult to compare individuals of radically different eras. I can see comparing
pitchers from the '40s with today's pitchers --my guess is that Bob Feller would probably have been a star today, because he was a fastballer. However, try to convince me that a pitcher winning 40 games a year and going nine inning in the majority of them was serving up anything more than a batting-practice fastball. Even though there may have been the occasional Ty Cobb or Rogers Hornsby at the plate, my guess is that the vast majority of major-leaguers back then were mechanics or farmers in the off-season.

Here's a simple EXCEL project -- compute a weighted average as follows: weight the best years of the pitchers in your list by the number of votes. For instance, if Robin Roberts' best year was 1950, multiply 1950 by 127. My guess is that this weighted average is less than 1920.

My immediate reply to him:

I don't know if I want to do that just yet (I might). But that center of gravity would be very, very different if we were to return to this project in about 7 years. Part of it is the failure of the 1980's generation to develop any candidates for being top-20 all-time. Oh, we elected some, like Stieb and Saberhagen, but they don't have the stature to stand up here. The generation of 1990's-centered pitchers is very different - but we haven't elected any of them yet because their careers were so very long that they're not eligible yet. (The first one up will be Kevin Brown, eligible for the 2011 election.)

I would think that all of Clemens, Maddux, Randy Johnson, and Pedro Martinez would rank in the top 20, with most of them in the top 10. And seeing a generation like that makes it a little easier to believe that Walter Johnson, Pete Alexander, and Joe Williams could have been contemporaries.

Part of this is the value v talent debate again. We don't have to answer the talent debate (which he seems to be making) because our charter is to recognize value.

The voter who omitted Feller was JPWF 13

Yep that was me, I had him 21-25 and I didn't give him WWII credit at all (for reasons which have been discussed.

My big outlier compared to everyone else was actually 3 finger brown...

I would have to agree that Blyleven above Koufax is a pretty incredible position to take.I wonder if the peak voters were underrepresented. I mean, Koufax was the top peak pitcher of the '60s if you look at more than a 1 year peak. He was the best pitcher in baseball for a solid 5 year period. Blyleven was none of these things. I've never thought of it in exactly these terms before, but if you were the manager and you needed to select a pitcher for a big game (7th game World Series, e.g.) after Jack Morris ;-) who would you take--Blyleven or Koufax. The answer to that is so easy it's silly.

agree w/ sunny; even tho he & I vote very differently, obviously if you set up the criteria to be weighted more toward the Q he just asked, Koufax is a no-brainer for being near the top. Others who view our project will have to understand many voters' perspectives before casting stones.

Even though there may have been the occasional Ty Cobb or Rogers Hornsby at the plate, my guess is that the vast majority of major-leaguers back then were mechanics or farmers in the off-season.

Maybe, but the question is how they would pitch if they had been born 20-30 years ago. For some reason, almost anytime we get these era comparisons, it's rarely factored in that the players of yesterday would be bigger, stronger and better educated in the ways of baseball play. Likewise, the stars of today would experience the opposite effect had they born years earlier. IOW, this argument doesn't really belong in an evaluation of the greatest players of all-time, IMO.

I've never thought of it in exactly these terms before, but if you were the manager and you needed to select a pitcher for a big game (7th game World Series, e.g.) after Jack Morris ;-) who would you take--Blyleven or Koufax. The answer to that is so easy it's silly.

...and if you were a GM with a crystal ball, you would have to be insane to pick Koufax over Blyleven at the start of their careers. ;-)

I do feel Blyleven is more of a mid-range HoM pitcher than a top-twenty one, FWIW.

I've never thought of it in exactly these terms before, but if you were the manager and you needed to select a pitcher for a big game (7th game World Series, e.g.) after Jack Morris ;-) who would you take--Blyleven or Koufax. The answer to that is so easy it's silly.

so what? And at what point in their careers?

and oh, Blyleven pitched extremely well in the postseason (not like Koufax granted)
and Blyleven pitched TWICE as much.

ages 26-30:


ye she had a great peak, so did some other guys:
ages 24-28


ages 28-32:


and ages 22-26


Koufax's "problem"- hes all peak
Take out that age 26-30 peak and you have 947 ip of 105 ERA+ ball
At the same ages that Koufax was throwing 947 IP of 105 ERA+ ball, Blyleven was throwing 1909 IP of 1332 ERA+ ball.

I think peak voters have been way OVER represented in the positional ballots, at least relative to the annual elections. I don't think it's been close.

Koufax/Blyleven are close here, 40 vs. 52 is nothing. There's no way it's a statistically significant difference.

Neither one of them were in the top 18, which is entirely reasonable. I'm not trading Phil Niekro's career for Sandy Koufax's. I can't imagine too many would. Sandy's in the mix after #18, where he belongs, unless you are a peak only voter. We've never had too many of those. We still ranked him in the top 1/3 or so of all HoM pitchers, I don't see how anything higher is defensible.

Can someone direct me to a thread discussion where actual methodologies to measure peak + career values might have been outlined ?

For example, would it make sense to look at a player's 5 year peak value and weight that 50%, and look at a players career value and weight THAT 50%, to come up with a more balanced approach to the peak vs. career debate ?

I'm sure this is not an original thought, so any help in locating discussions that deal with this issue specifically would be appreciated.

I do agree that our results were a little too skewed to the early years. I mean the top 4 and 8 of the top 10 before WWII? Seems to me there might have been an environmental thing that made it easier to rack up a long career at a high level during that time . . . I've always tried to separate players from their era when going across time.

I've never understood the support for Spahn. Obviously he's got the wins, but this community knows to look deeper. He's got the career length, but so does, say, Blyleven. In fact, here's a brief stat line:

6. Spahn - IP 5243.2, ERA+ 118
19. Blyleven - IP 4970, ERA+ 118

And yet people are questioning whether Blyleven is too high! Seems strange to me.

I found This article that addresses my question.

I took a quick look at Heyyoo's found article. The main thing that leapt out at me was using the average as the zero point, which the author appears to claim is what allows his method to properly balance peak and career. Well, if you want to do that, you really should start with Total Baseball's rankings. Pete Palmer, who is an inner-circle Sabermetrician of Fame by any standard, has spent the last 40 years perfecting a method with the average being the zero point, using Linear Weights as his statistical foundation tool. His rankings have similar strengths and weaknesses to the cited article, but have a much more thorough development. - Brock

Sandy Koufax should be water over the dam here.

17. Joe Dimino Posted: October 16, 2009 at 12:53 PM (#3354044)
I think peak voters have been way OVER represented in the positional ballots, at least relative to the annual elections. I don't think it's been close.

Koufax/Blyleven are close here, 40 vs. 52 is nothing. There's no way it's a statistically significant difference.

Yes to the latter.

Re the former, note that Blyleven (#6) ranks far ahead of Koufax (#11) in Pitchers 1958-1984+ (Group 4). Beside the bodies they are separated by one notable gap above Palmer & Jenkins, another below them. There Koufax is insignificantly behind Nolan Ryan, who played forever and didn't get a mention here; and Juan Marichal who didn't get a mention here either.

(Palmer & Jenkins were mentioned here but they trail Koufax by more than he trails Blyleven, all beyond the fringe where it is reasonable to cite the standings for the purpose of the exercise, the group ranking of pitchers overall. --that is, down in the realm of insignificance.)

18. heyyoo Posted: October 16, 2009 at 12:57 PM (#3354050)
Can someone direct me to a thread discussion where actual methodologies to measure peak + career values might have been outlined? ...

I think you mean balance peak and career or weigh them on the same scale, rather than measure them separately. That's an important question. There is some valuable discussion here (search the Archives?) but, ultimately, balance or weigh is what the election does.

I'm sorry, what was the question?

I have him for 6-7 great years and I have a chance to win. I keep Bert I get 15 good years and never have a better than average chance of a pennant.

Koufax had a higher peak than Blyleven no doubt, but people tend to overrate Koufax's peak
25-5 1.88
19-5 1.74
26-8 2.04
27-9 1.73

stunning, absolutely stunning right?
BUT
1: He did that in a VERY depressed run environment, Dodger Stadium in the 1960s
2: He did that for very good teams

1947-2009:

He appears 3 times on that list, his 27 win season is the top, his 25 win season tied for 2nd
Since 1947, pitcher has won 25+ games 21 times, Koufax did it 3 times (matched by Marichal)
1947-2009, top ERA+ (200+ IP)

He appears once? If you expand the list to include all of Koufax's sub 2.00 ERA you get 100 seasons. What happened? Adjusting for park and league you see that Koufax's peak is up there with guys like Seaver and Clemens and Maddux and Martinez and Gibson and Randy Johnson, but he doesn't stand high above them (if at all) and they all had longer and more effective careers outside their peaks than he did.

Plus or minus a few spots, most of the ranking seems about right - or at least reasonable to me. But what I can't figure out is how Robin Roberts came in 16th whereas his virtual clone, Fergie Jenkins, came in 35th.

Their career stats are probably closer to identical than any other two pitchers on this list, with the sole outlier being that Jenkins struck out significantly more batters. So what's the story? Was Roberts' peak that much better?

Was Roberts' peak that much better?

Yep.

Obviously ERA+ and IP aren't everything, but I still like IP*(ERA+ - 80) for a quick and dirty estimate of value above replacement. That's probably a generous baseline, so that helps long career guys like Spahn. And the top 25 are

1. Cy Young 427
2. Walter Johnson 396
3. Roger Clemens 310
4. Kid Nichols 303
5. Pete Alexander 285
6. Lefty Grove 268
7. Christy Mathewson 263
8. Greg Maddux 260
9. John Clarkson 245
10. Tim Keefe 237
11. Randy Johnson 232
12. Tom Seaver 225
13. Pedro Martinez 209
14. Warren Spahn 199
15. Gaylord Perry 198
16. Ed Walsh 196
17. Phil Niekro 189
18. Bert Blyleven 189
19. Eddie Plank 189
20. Amos Rusie 185
21. Mordecai Brown 184
22. Steve Carlton 183
23. Bob Gibson 183
24. Jim Palmer 182
25. Carl Hubbell 180

This list is about as career-length friendly as you can make, and Spahn just doesn't come out as the super elite (nor Carlton or Gibson, whom most don't lump with Palmer). To put it another way, Spahn only had two seasons in his career better than, say, Sabathia in 2009. Only three seasons clearly better than Randy Wolf in 2009.

Was Roberts' peak that much better?


Yep.

Well then I guess that it 's not just a question of peak vs. career, but also how do you define "peak".

Roberts basically had a six year peak that was easily better than any 6 year peak that Jenkins had. But Roberts was done as an elite pitcher at age 28 - he was basically just about league average from that point on. And if not for a second mini peak at age 35-38 he would have been well below average.

Jenkins peak, though not as high, lasted twice as long. Instead of making a comeback at 35, that's when he started to slow down. And even then, he was still just about league average for his last 5 seasons. Unlike Roberts, he never actually hurt his team by taking the ball 40 times a season.

So I'm curious, if the emphasis is on peak, then why isn't "equal credit" given to troughs?

I think people who like to consider peak value, a peak has to be about three years to get our attention. But other than that, it's just a question how high it is. You can probably thank Bill James for that, what with his three year peak in the BJHA. So yes, Roberts peak was better.

30. Ben V-L Posted: October 17, 2009 at 04:12 PM (#3355231)
I still like IP*(ERA+ - 80) for a quick and dirty estimate of value above replacement. That's probably a generous baseline, so that helps long career guys like Spahn.
[list deleted]
This list is about as career-length friendly as you can make, and Spahn just doesn't come out as the super elite (nor Carlton or Gibson, whom most don't lump with Palmer).

(nor Seaver) Does "Spahn come out as super elite" in the official ranking? Both slot him third behind Grove and Seaver among all pitchers between Alexander and Clemens, more than seventy years later. Call it the 1920s to 1980s. The differences pertinent to Spahn are that IP*(ERA+ - 80) puts seven pre-1920 pitchers and four post-1990 pitchers before & with that trio, instead of three and none.

Between Alexander and Gibson, almost fifty years later, IP*(ERA+ - 80) slots only Grove and Spahn among the 24 leaders.

Jenkins peak, though not as high, lasted twice as long.

Yeah, as Sunnyday said, that means to us that he had a lesser one. "Not as high" is way more important to a peak voter (once you get over the three years or whatever our personal minimum is) than "twice as long." (Unless, of course, it isn't "not as high," in which case twice as long at the same high level is, well, twice as impressive.)

The definitions blur, and it's all semantics in the end, but I think I'd even call what you're calling Jenkins's peak his "prime."

As for troughs, not sure what you mean. Do you mean deducting credit for lousy years? Most of us have a philosophy that someone being trotted out there when he's performing below replacement level is more on the manager/team than on the player, and we don't believe in negative credit. (Not all of us, though.)

It's a different point, regardless: Peak voters feel that we want, in varying degrees of intensity depending on taste, to be considering and honoring the players whose highest level of performance was higher than most everyone else's in history. Again, there's a minimum-length requirement on that--I'm not voting for Dwight Gooden, though that one fantastic season does get him more consideration from us peak types than from most--but once that's met, it's all about the heights of performance the player reached. Lousy seasons and troughs have no effect on that.

And, as I always say at the end of these kinds of posts, perhaps the best explanation I can give is this: I vote for Al Rosen in every election. :)

Without my enthusiastic (misguided?) support of Rogan it looks like he might not have been in the top 30.

I wanted to chime in about Spahn, is he just towers in wins over the guys who pitched at the same time as him.

Spahn is #1 in wins for the last 98 seasons inclusive.

As for troughs, not sure what you mean. Do you mean deducting credit for lousy years? Most of us have a philosophy that someone being trotted out there when he's performing below replacement level is more on the manager/team than on the player, and we don't believe in negative credit. (Not all of us, though.)

What I mean is that at age 29 Roberts basically fell off a cliff and managed only one 100+ ERA+ over the next six seasons. In what should have been his prime years he didn't decline so much as he vanished into a black hole. By way of contemporary comparison he was basically Shawn Estes at the same age with one good season added in. Only a couple of good seasons from age 35 onward saves his reputation and gets him into the HoF and HoM. Jenkins in contrast never had that black hole and pitched without much discernable decline for at least a dozen years. Put their statistics side by side until age 35 and I find it hard to see Roberts adding more value. In fact, if Roberts didn't have a revival in Baltimore he doesn't make this discussion.

BTW, I'm not trying to raise Jenkins's ranking so much as to understand Roberts's.

Let me put it this way: I find it hard to see how anyone could rank Robin Roberts ahead of Sandy Koufax. Koufax far and away beats him on peak, and Koufax's first 6 years were far and away better than Roberts's 29-34 age period. Do Roberts's 3 good but less than exceptional years after moving to the AL really place him ahead of Koufax? I just don't see it.

So ... if you're a peak voter, how do you put him ahead of Koufax? And if you're a career guy, how can you not pair him with Jenkins?

I have a hard time seeing that 12-season "peak" or "prime" by Ferguson Jenkins. He declined sharply in Chicago. The Cubs and Rangers gambled on him and Bill Madlock (OPS+ 171 in 21 games, September 1973), a winning gamble on the Cubs side. Jenkins famously won 25 games next season, a great season but nothing like Gaylord Perry's GREAT 1972 debut on the junior circuit (altho I equated them at the time).

Of course, Jenkins was always much better than "Robin in the Trough". Regarding the big question, based on previous special elections and their discussions I predicted that Blyleven, Palmer, and Jenkins would all win spots in the top twenty here, while Marichal, Koufax, and Ryan would enjoy support (that is, 8 to 11 from group four). What happened, I suppose, is that voters working down to Palmer, Jenkins, etc, on their lists --as for Plank, Walsh, and Brown from the 19-aughts-- discounted their fine raw and relative statistical records as some reflection of easy conditions for pitching. The sheer number of contemporaries on the ballot and on the bubble seemed incredibly high. That may be reasonable. The chronological group elections were misleading, however, because within the narrow confines of "1959 to 1984+" the voters were comfortable saying essentially that all of the pitchers from the first half were greater than all of the pitchers from the second half.

Robin Roberts actually overlapped in the major leagues with the 1960s generation including Palmer & Jenkins. If the 1960s conditions were uniquely good for young pitchers, however (the prevailing view around here), that didn't help Robin Roberts on a career scale. He did make his mid-thirties comeback in the 1960s. He was a high fastball pitcher, unusually vulnerable to the home run. Maybe the high strike zone of 1963-68 helped him more than most. The comeback clearly occurred in 1962, however, his AL debut season with Baltimore, one year before the rules change. That seems decisive.

The outstanding feature of Robin Roberts at his peak was his heavy workload in combination with excellent rates. By innings pitched his margins over the third-leading NL pitcher were roughly 5%, 5%, 20%, 35%, 30%, 20%, 5% (his rank, 2 1 1 1 1 1 2).

while Marichal, Koufax, and Ryan would enjoy support

I'm pretty sure someone will dismiss this whole exercise because 36 pitchers got a vote and Nolan Ryan wasn't one of them. They would have a point - Ryan is in my top 36 all-time. He's just not in the top 20.

I have a hard time seeing that 12-season "peak" or "prime" by Ferguson Jenkins. He declined sharply in Chicago.

Jenkins won the CYA with his best season in 1971. In 1972 he went 20-12 with an ERA+ of 119. He came in third in the CYA voting in his 6th straight 20 win season. In 1973 he "bottomed out" with an ERA+ of only 101, and in a panic the Cubs traded him. So yeah, in that context you could say that "he declined sharply in Chicago" from 1971 to 1973.

But of course if you look at the next few seasons in the AL you'll see that basically he continued on pretty much the same level of performance with just one more off year thrown in. Starting in 1967 he posted the following ERA+ marks: 127, 120, 126, 131, 142, 119, 101, 126, 95, 121 124, and 124. Yes, two of those years were pedestrian, but unless you want to narrowly define his peak as 1970-1971, it's hard to argue that he didn't pretty much maintain a consistent level of performance for those 12 years. So what don't you see?

And yes, I know that ERA+ is not the be-all, end-all barometer of a pitcher's season. But as a quick and dirty indicator it's good enough.

The outstanding feature of Robin Roberts at his peak was his heavy workload in combination with excellent rates. By innings pitched his margins over the third-leading NL pitcher were roughly 5%, 5%, 20%, 35%, 30%, 20%, 5% (his rank, 2 1 1 1 1 1 2).

If you're trying to differentiate between Roberts and Jenkins, you've definitely cherry picked the wrong stat. First of all, you've included his 1956 rank where he pitched almost 300 innings with an ERA+ of 84. Naughty, naughty.

Jenkins in comparison ranked 3, 2, 5, 2, 1, 2, 5, and 2 in IP from 1967 to 1974. Admittedly not quite as impressive as Roberts, but for a longer stretch of time.

However, to merely compare rankings ignores a significant difference between the eras that they pitched in, namely their competition. A quick check of the IP leaders in AL 1950s shows that Roberts was basically competing against Spahn and a whole lot of other guys that didn't make it to this discussion thread. Jenkins, on the other hand, was pitching in the NL's golden age of pitching and was getting the ball as much if not more than fellow ranked pitchers such as: Seaver, Gibson, Carlton, Perry, and Niekro. And some of the other guys near the top of the lists were named Marichal, Drysdale, Ryan and Sutton. Jenkins was a standout workhorse in an era of workhorses. If you're giving credit for providing high quantities of quality innings, Jenkins gets as much if not more than Roberts.

Jenkins was a standout workhorse in an era of workhorses.

Gaylord Perry says hi. From 1967 to 1974 Perry ranked 2, 4, 1, 1, 4, 2, 2, and 3 in IP for a total of more innings over that same 8 year stretch plus at a higher ERA+. Can you stand out when you're behind another guy?

If you're giving credit for providing high quantities of quality innings, Jenkins gets as much if not more than Roberts.

Roberts condensed his high quantities with slightly higher quality innings in his peak. Combined with the fact that the 60s/70s are overflowing with quality pitchers compared to the 50s it's plausible to place Roberts over Jenkins.

It isn't clear how one should account for the so-called competition.

Let me rearrange.
Within the narrow confines of "1959 to 1984+", it appears, voters were comfortable saying essentially that all of the pitchers from the first half were greater than all of the pitchers from the second half. In the general election, it appears, voters discounted their fine raw and relative statistical records of the same as some reflection of easy conditions for pitching. The sheer number of contemporaries on the ballot and on the bubble seemed incredibly high.

This exercise essentially identified 18 outstanding pitchers --Clarkson, six from chronological group two, six from group three, and five from group four. From the major leagues under 60'6", we identified fifteen --five, five, and five.

To put it another way, Spahn only had two seasons in his career better than, say, Sabathia in 2009. Only three seasons clearly better than Randy Wolf in 2009.

Spahn had just three seasons "clearly better" than a 214 IP, 129 ERA+ season?
By your quick and dirty metric (era+-80*ip) he had 12 seasons that were better
Wolf has finished in the top 10 in ip ONCE, 2009 he was 10th
Spahn finished in the top 5 a total of 14 times, he was 1st or 2nd in IP a total of 10 times
302 ip with a 124 ERA + is most definitely better than 129 in 214 ip, even allowing for era.

Jenkins was a standout workhorse in an era of workhorses.

Gaylord Perry says hi. From 1967 to 1974 Perry ranked 2, 4, 1, 1, 4, 2, 2, and 3 in IP for a total of more innings over that same 8 year stretch plus at a higher ERA+. Can you stand out when you're behind another guy?

Good catch on Perry. But to answer your question, yeah sure you can be a standout even when someone else is better than you. Just for fun, go rank the top 5 contemporary pitchers expected to become HoF eligible over the coming years. Wouldn't you consider them all standout pichers?

If you're giving credit for providing high quantities of quality innings, Jenkins gets as much if not more than Roberts.

Roberts condensed his high quantities with slightly higher quality innings in his peak. Combined with the fact that the 60s/70s are overflowing with quality pitchers compared to the 50s it's plausible to place Roberts over Jenkins.

Perhaps, but flipping the logic around makes just as much sense: Combined with the fact that the 60s/70s are overflowing with quality pitchers compared to the 50s it's plausible to place Jenkins over Roberts. You could easily make the argument that Roberts benefited by comparison to his less than outstanding peers. Had he been a 60s/70s pitcher his normalized stats may have been quite unremarkable. We'll never know.

In any case, my original point was that I found it hard to rate Roberts that far ahead of Jenkins since they have such similar stats. As it is, the more I look at it, the more flabbergasted I get that anyone would rate Roberts in the top 20. Seriously, the guy had a career ERA+ of 113 (244th all time) and was a league average pitcher from the age of 28 onwards. And it's not like his rates went down because he hung around too long - if he hadn't had a turnaround at age 35 it'd have been worse - seriously, I doubt he makes the HoF if he retired at 35. I can't see why any of the later pitchers on the list would be ranked behind him.

Also, it seems rather arbitrary to downgrade 60's pitching simply because there were a lot of good ones at the time. I'm sure that most would also downgrade contemporary hitting, but I suspect that no one would think to do likewise for NL hitters - especially HR hitters - from the 50's. Mays, Aaron, Mathews, Banks and even Stan Musial benefitted from the richest HR environment in history up until the 90's. Anyone care to discount their stats?

Perhaps, but flipping the logic around makes just as much sense: Combined with the fact that the 60s/70s are overflowing with quality pitchers compared to the 50s it's plausible to place Jenkins over Roberts. You could easily make the argument that Roberts benefited by comparison to his less than outstanding peers. Had he been a 60s/70s pitcher his normalized stats may have been quite unremarkable. We'll never know.

Since it was easier on the pitcher per season and on a career basis during the '60s/'70s, the odds are pretty good that Robert's career would have been only better, not worse. That's why Roberts belongs over Jenkins, IMO.

Average season in six-year prime

So ... if you're a peak voter, how do you put him ahead of Koufax? And if you're a career guy, how can you not pair him with Jenkins?

Because no one here is a nothing-but-peak or nothing-but-career voter, I'm pretty sure. Even for us extreme-peak guys (I'm among the most extreme), it's a mix. Koufax's peak is wonderful, but so short that anyone with a somewhat lesser one but something resembling a solid career (which, despite the troughs, I think Roberts has) can pass him in my system. By the way, for an extreme-peaker like me, he doesn't pass him by much: I had them at 15 (Roberts) and 16 (Koufax) on my ballot.

And on the other hand, I still do weigh peak pretty heavily, so Jenkins doesn't get into my top 20, and Roberts's peak IS enough to get well past him for me.

<cough>

21-25) Rusie, Roberts, Plank, Ray Brown, Jenkins

Koufax's peak is wonderful, but so short that anyone with a somewhat lesser one but something resembling a solid career (which, despite the troughs, I think Roberts has) can pass him in my system.

Put Koufax's career side by side that of Roberts first 12 seasons. There will be a bit of a lookingglass effect: Koufax starts out slow and finishes with a bang, while Roberts starts strong and then drops off. To make the comparison easier, list the seasons for each of them in order of best to worst using whatever methodology you fancy. Koufax is clearly the far superior pitcher to this point in their careers; his peak is of equal length and higher, and the rest of his seasons match up well enough with the beginning of Roberts' post age 28 trough.

From this point on Roberts pitched 7 more seasons. 2 of them were garbage; 2 of them were basically league average; and the remaining three represent his late career revival. So assuming that league average and under seasons don't add or detract much to/from "a solid career", I guess you're saying that those three good late career seasons were enough to wipe out Koufax's substantial lead. I don't see it.

Ok, here's Koufax's first 12 years, ranked by (ERA+ -80)*ip ("
score")

The total is 147,421
Here's Robert's first 12 years:

total is 140,307
Outside of those 12 years Roberts also has these years:

Of curse ERA+ -80 use an 80 ERA+ as replacement level, personally, when talking about HOM I would use a replacement level of 100, in which case, Koufax wins, even if you exclude all the negative seasons

Well, I compare top seasons to top seasons, regardless of chronology, as you suggest. And I use various metrics, but lean heavily on BP's PRAA. In that, Roberts is only marginally behind Koufax after the best 7 years of each player, and has caught and passed him after the best 10. (Sorry, no time to drop a chart in here right now.) I guess that does mean I feel those late-career season do the job, but it's also that Koufax's lead isn't really all that substantial (by the metrics I'm using, anyway) once you get past the top 5 or so seasons for each.

However, because I weigh those peak seasons more heavily, the fact that Roberts is past Koufax after the top 10 seasons for each doesn't actually get him by Koufax in my system. The rest of Roberts's career doesn't quite do it either (it's not that great anyway, as you say). I should have added before that I have adjustment for 50s pitchers, as they consistently do poorly in all these metrics, and I don't buy that Warren Spahn was only a moderate HOFer and Whitey Ford wasn't one at all. The adjustment I use to fix that also boosts Roberts--in this case, just enough to get by Koufax. (It's still pretty close.)

Which means, of course, that you're right: without that adjustment, Roberts can't get past Koufax's early lead. But I also think we disagree on how substantial that lead really is.

Also, I fall on John Murphy's side of the comparison-to-peers argument. That of course has more to do with the Jenkins comparison than the Koufax one.

Here's my methodology on that. That is, RA+ Pythpat, computing equivalent records by assuming 1 decision for each 9 IP, then computing equivalent Fibonacci Win Points for each season. I then sorted each pitcher's career by those equivalent FWP.



So this shows Koufax's top 4 years as being better than Roberts's top 4 years - but it really isn't all that big an advantage. And then the prime-shoulder seasons for Roberts start kicking in, and go on for a long ways. Year 5 already makes up about a third of the difference in the top 4.

That the top end is so close is a bit of a surprise. I suspect that's mainly from the sliding exponent adjustment - Koufax had his peak in lower scoring times than Roberts and hence needed a better RA+ for the same winning percentage. There may also be a lesser effect from Roberts allowing less than his share of UER, although that's probably not big.

What things does this, RA+ Pythpat miss? The most important is defensive support. I don't have a very clear picture on that as a comparison between these two; I suspect that they both had fairly good defensive support at their peaks. There may have been a particular synergy between Roberts and Richie Ashburn where the sum was greater than the parts.

The second most important thing is the pitcher's own offense - and there is a difference there. It's not that Roberts was anything special as a hitter (career OPS+ 27, 108 sacrifices in 1782 PA) but that Koufax was a historically awful hitter (career OPS+ -26, only 35 sacrifices in 858 PS so he wasn't even much of a bunter). Make an appropriate deduction in the above for Koufax, and the top four years draw even closer.

This conversation started in part about Roberts versus Jenkins. On that: yes, Roberts had a higher peak, but Jenkins makes quite a bit of it up on the back end. I had my vote as 15 Niekro, 16 Perry, 17 Roberts, 18 Plank. 19 Blyleven, 20 Palmer, with Jenkins just outside the top 20; I could easily see rearranging that list to move Jenkins up.

FWIW, Roberts had 91 RC in 1422 outs, Koufax had 16 in 750 outs
Among pitchers with 800+ PAs, Koufax's .261 OPS is 185th out of 186. His OPS+ of -26 is also 185th out of 186.

Roberts was a pretty average hitter for a pitcher
Koufax was execrable as a hitter, so much so that his hitting was about a 4-5 win difference between him and Roberts.

If you drop the PA cutoff to 502, Koufax is 476th out of 488, btw, wow was Dean Chance a terrible hitter.

Of curse ERA+ -80 use an 80 ERA+ as replacement level, personally, when talking about HOM I would use a replacement level of 100, in which case, Koufax wins, even if you exclude all the negative seasons

Then we are in total agreement.

btw, wow was Dean Chance a terrible hitter.

It's a little surprising that Chance drew as many walks as he did: 30 career walks as compared to 44 career hits. I suppose that's what happens when you seldom make contact - since the PA doesn't end on contact, the pitcher does have to put that many pitches into the strike zone. Koufax had 43 walks and 75 hits. The average-ish Roberts had a BB/H ratio of 135/255. For Spahn, it was 94/363. I have noticed that pitchers recognized as good-hitting pitchers (a category that might include Spahn) still tend to draw very few walks.

Yeah, Chance looks like someone who might have batted 9th on his high school team.

And since I commented on Koufax's sacrifices: Chance had 61 in 759 PA. Koufax's sacrifice rate is quite strikingly low, especially since I can't think of any good motive for not calling for the bunt whenever possible for such a hitter.

Koufax's sacrifice rate is quite strikingly low, especially since I can't think of any good motive for not calling for the bunt whenever possible for such a hitter.

Two theories:

1) Over half of Koufax's PAs were from 63-66 when the Dodgers had a pretty low scoring team even for a low scoring era. Perhaps there just weren't a lot of guys on base when he got up.

2) Koufax was such a crappy hitter - about half of all of his ABs ended in strikeouts - he probably had a hard time getting bunts down. I suspect that he just had a really low success rate.

Besides, how complete a game do you expect from a guy that went to college on a basketball scholarship?

Since it was easier on the pitcher per season and on a career basis during the '60s/'70s, the odds are pretty good that Robert's career would have been only better, not worse. That's why Roberts belongs over Jenkins, IMO.

I subscribe to the theory that in extreme league conditions absolute comparisons of normalized rate stats become less reliable.

Put differently, as league ERA approaches zero, it becomes increasingly difficult to get "separation" from the pack - there's only so low that you can go. In statistical terms the lower the average stat, the smaller the standard deviation and the closer outliers get to the mean. By way of example, Gibson had an ERA+ of "only" 258 in 1968, while Martinez posted a 291 in 2000. In absolute terms Gibson's ERA+ is obviously less impressive, but I'm not convinced that it indicates a lesser accomplishment.

In this context, I don't penalize Jenkins for supposedly pitching in an era "was easier on the pitcher per season and on a career basis" - I give him extra credit. An ERA+ of 125 in 1965 sounds more impressive to me than an ERA+ of 125 in 1955 - or 2005 for that matter.

And btw, I also don't have any problem saying that some eras just have better/worse overall/individual hitting/pitching than others. The 60s and 90s just happened to have more great pitchers than the 50s. I'd be more surprised if there was an even distribution across all eras.

And btw, I also don't have any problem saying that some eras just have better/worse overall/individual hitting/pitching than others. The 60s and 90s just happened to have more great pitchers than the 50s. I'd be more surprised if there was an even distribution across all eras.

I don't disagree with this. The '60s/'70s did have better pitchers than the '50s. Still doesn't mean that Fergie was better than Roberts, however, since he wasn't. ;-)

One other thing: Roberts averaging 323 innings per season during his prime is much more impressive than if he had during the '60s/'70s.

Apparently, the Standard Deviation for ERA+ was lower in Roberts' prime.

From 1950-55 there were 4 pitchers with ERA+ of 160+ or 1 for every 24 teams (96 teams) (215 pitchers with 190+ IP)
From 1961-66 there were 13 pitchers with ERA+ of 160+ or 1 for every 9.1 teams (118 teams) (254 pitchers with 200+ IP)
From 1967-72 there were 16 pitchers with ERA+ of 160+ or 1 for every 8.5 teams (136 teams) (332 pitchers with 200+ IP)

This is what one would expect to happen, with Roberts playing in a contraction era (NeL demise) and Koufax and Jenkins in an expansion era.

Put differently, as league ERA approaches zero, it becomes increasingly difficult to get "separation" from the pack - there's only so low that you can go. In statistical terms the lower the average stat, the smaller the standard deviation and the closer outliers get to the mean.

The basic argument doesn't pertain to "relative statistics" which approximately or literally compare raw values to average ones. ERA+ is one example.

The basic argument pertains to raw non-negative measures. --and to measurements with a ceiling such as 100% score or 800 SAT math.

The standard deviations generally need to be calculated, not assessed by basic argument.

Responding to LSR's post 59:

I looked at the career sacrifice rate for a number of 60's pitchers. Here's a table; the number on the far right is PA per SH, so a higher number means fewer successful sacrifices.


Earl Wilson: 22/838, 38.1
Koufax: 35/858, 24.5
Podres: 43/853, 19.8
Drysdale: 69/1309, 19.0
Marichal: 71/1339, 18.9
Ford: 65/1208, 18.6
Osteen: 69/1220, 17.7
Bunning: 85/1401, 16.5
Chance: 61/759, 12.4
McClain: 64/709, 11.1

(You could probably learn more from a bigger sample.)

There might be a little of LSR's reason #1 - something about the Dodgers bunting less than most teams (the position of Podres on the list is part of why I say that) - but I still think it's mostly reason #2.

The general tendency is for bad-hitting pitchers to bunt more. The three worst hitters on this list are Koufax, Chance, and McLain, and Chance and McLain each bunted a lot. Earl Wilson is an extreme outlier, but he's also the best hitter on this list, with a lifetime 76 OPS+ and .174 ISO. If he comes up with runners on first and second, you'll be thinking about the three run HR/.

Maddux and Johnson both broke the record best "winning percentage" for starting pitchers in 1995.

The schedule was 11% short of 162, which increased the probability of very high winning percentages for that season. What else? Maddux started 28 and Johnson 30 games.

W-L Pct. GS (CG,inc) GR dec/gs win/gs
19-2 .904 28 (10,18) 0, .750 .678, Maddux 1995
18-2 .900 30 ( 6,24 ) 0, .667 .600, Johnson 1995

25-3 .892 35 (16,19) 0, .800 .714, Guidry 1978

31-4 .885 30 ( 27,3 ) 11, 1.16 1.03, Grove 1931

28-6 .823 35 ( 30,5 ) 0, .971 .800, Hughes 1899

(continued)

Previously Ron Guidry 1978 finished 25-3 in 35 starts, 16 complete. Maddux & Johnson surpassed his official winning percentage (column two) but they earned decisions at lower rates which their higher winning percentages did not overcome; they earned wins at a lower rate per start (two bold rates).

Looking further back, Lefty Grove 1931 finished 31-4 in 41 games including 30 starts (27 complete) and 11 reliefs. He earned more decisions than starts, also more wins than starts (bold).

Jay Hughes 1899 seems to be the standout season by winning percentage among the relatively few pure starting pitcher seasons from long ago. He finished 28-6 in 35 games, all starts, 30 complete. Same as for Maddux and Johnson compared with Guidry, for Guidry compared with Hughes, the increase in winning percentage (column two) is too small to counter the decrease in proportion decisions/starts. There is a net decrease in proportion wins/starts (bold).

Highest Win%, 20+ Dec, 1893-2009

Jay Hughes 1899 seems to be the standout season by winning percentage among the relatively few pure starting pitcher seasons from long ago.

Bill Hoffer was 31-6, .838 in 1895. He has the only pct higher than Hughes in the period 1885-1900.

Hoffer pitched three games as a relief pitcher where I looked for pure starting pitcher-seasons so that the season record would be the record as a starting pitcher (because I don't have a handy source for start and relief pitching records). Looking up his season-by-season record at bb-ref now, I see that Hoffer 25-7 .781 in 1896 pitched 35 games, all starts. That's a likely candidate for second-best W-L to Hughes among the early pure starting seasons.

The Orioles remained a great team, which certainly helped Hoffer as he declined to league average, ERA+ 97 but W-L 22-11 in 1897.

John at bat,
The rank "RK" should run to 37. and 38., not 35. and 36. (Thanks to a nonymous observer off-site.)

Of curse ERA+ -80 use an 80 ERA+ as replacement level, personally, when talking about HOM I would use a replacement level of 100

Then, by this metric, Spahn is ranked 26th for career. And career length is one of his strengths.

A minor comment: ERA+ uses pitcher ERA as the denominator, and as such is non-linear in pitcher's ERA. This makes IP*(ERA+ - 80) a somewhat misleading stat, even as a sort of rough measure. In particular, it tends to overrate pitchers with very high ERAs but comparatively short careers; look at how high Pedro is in post #30, compared to where he normally ranks in various types of WAR.

Mister Secretary-General,
This and a few of the preceding pitcher rankings threads need listing in the "Important Links" subsection for elections.


In the choneWAR thread DL from MN posted "Top unelected pitchers from Chone" (pitching only). I revised that to rank pitchers overall. See Chone's WARP and the Hall of Merit, #87 and #32. DL closed, 'Why aren't the "Chone Voters" filling their ballots with pitchers? His system is saying they're more valuable than the hitters they're voting for.'


Here is the Combined Pitchers ballot of straight Chone Voters (certainly not choneWAR author Sean Smith).

1. Cy Young 143.4, the big leader by pitching

2. Walter Johnson 139.9, a very good batter who played a lot for a long time, but not quite enough to match Cy Young's sheer pitching.

(huge gap)

3. Pete Alexander 106.6
4. Tom Seaver 106.1
essentially a double tie at pitching 105 and moderate peripherals, far behind the Big Two in sum. Is it a bigger gap than the shortstops behind the Flying Dutchman?

5. Kid Nichols 103.3, another with moderate peripherals

(big gap)

6. Lefty Grove 93.4
7. Phil Niekro 93.1
8. Bob Gibson 91.6, by very good batting he surpasses Matty (good) and the next three (poor)
9. Christy Mathewson 90.8
10. Gaylord Perry 90.5, by pitching alone ranks 8 just behind Niekro; a below average baserunner and one of the poorest batters
11. Warren Spahn 89.7, why does he have any kind of reputation as a batter? is it nothing but homeruns? Spahn couldn't bat with Blyleven and he was a poor baserunner.
12. Bert Blyleven 87.6, weak batter
13. Steve Carlton 87.1, strong batter, just enough to make a baker's dozen elite pitchers

(gap)

14. Robin Roberts 82.6, well above average at bat, enough to move four rungs up the ladder
15. Ferguson Jenkins 81.5, it's all hugely underrated pitching; overrated for batting thanks to a half dozen homeruns
16. John Clarkson 81.2, it's all pitching
17. Nolan Ryan 80.6, very weak batter whose pitching matches Carlton
18. Tim Keefe 79.2, very weak batter whose pitching matches Clarkson
19. Eddie Plank 76.9, it's all pitching

(gap)
20. Charley Radbourn 73.6, very good batter for a pitcher who sometimes played the outfield; poor baserunner and fielder

Bob Caruthers 71.4, always outclassed the pitchers at bat, briefly a league leader. For two seasons in the glory days of Comiskey's St Louis Browns, "Parisian Bob" was a regular outfielder when not on the mound; teammate Dave Foutz was another. Sold to Brooklyn for 1888, he kept the dual regular role for one season and helped cover the outfield thereafter. In 1992 he returned to St Louis for one season as everyday outfielder. Career OPS+ 133.

Don Drysdale 69.7, very good batter with homerun power, occasional pinch-hitter or -runner although a poor baserunner.

Tony Mullane 69.5, super batter among pitchers, occasionally played elsewhere. Mullane is not in the Hall of Merit.

No minor league credit for Grove in that list.

(Sorry, Mets fans. Tom Seaver should be boldly named at rank #4. I now how it happened, never mind.)

What DL says. For the rankings and ratings I have revised the published "Top 500 Pitchers" (by career pitching choneWAR) in one way only: add career batting/running/fielding choneWAR and re-order by the sum. For most of the comments I have simply referred to the components of batting/running/fielding choneWAR at the career level, which Sean Smith provides on the player pages linked to the Top 500 listings.

Any adjustment in the rating is likely to disturb the ranking because so many of the rating differences are tiny. For example, with pro-rata credit for summer 1981 and no other adjustments Seaver, Niekro, Perry, Blyleven and Carlton would climb one rung each; Nolan Ryan would climb three rungs; Fergie Jenkins would drop below Clarkson. (Seaver, Blyleven, Carlton, and Ryan enjoyed excellent split seasons.)