Ok, I'll move my comment over here...

OCF,

Taking pitching holistically over the eras, lets jump ahead 20 years and see where we are in 1935 (the year Eddie Plank comes on).

I think, looking ahead, you've got two categories of pitchers.

Group One are the definite #1s (or maybe #2s, depending on who else is on the ballot . . .): Cy Young, Pete Alexander, Christy Mathewson, Walter Johnson, and Eddie Plank.

Add those 5 to the 6 already in, and you've got 10 pitchers through 1935. (10 1/2 with Ward). By 1935, we will have elected 65 persons total. That leads to a HoM of about 17.5% pitchers. Assuming that people generally have a "non-quota quota" of between 1/5 and 1/3 pitchers, then electors will want to elect between 2 more (to get to 20%) and 10 more (to get to 33%) pitchers.

Group 2 is everyone else.

The question will be, will we elect closer to 2 or 10 additional pitchers, and which pitchers will those be.

I personally think 2 is way too low. We'd end up with an all-superstar pitching roster, compared to a broader range of "just great" hitters. I'd be stretching, though, to find another 10 I found worthy, even though I fall theoretically more on the 33%-pitching side.

Er, that's:

"Add those 5 to the 6 already in, and you've got 11 pitchers through 1935. (11 1/2 with Ward)."

The rest of the math is right.

MattB:

I'm guessing you meant to say "1935 (the year Pete Alexander comes on)". Plank is a decade or so before that.

Of course I agree with you. There's going to be no serious disagreement about Young, Mathewson, Johnson, or Alexander, but that won't be enough pitchers. That's exactly why I asked the questions above - which other pitchers?

Meanwhile, here are seventeen of the top Category 2 pitchers -- of which we will probably admit between 2 and 10 -- arranged by era. I'd be happy if two or three or each group were admitted.

19th Century Pitchers:

Big Ed,

I screened pitchers with 200 or more wins (which is, essentially the list of the top 100 in career wins). I then took out the pitchers who haven't gotten any support (e.g., Gus Weyhing). Your 195 fell a little short, essentially because your career was only 7 years long. Is 80% of Amos Rusie enough? Maybe to the voters, and Walsh is certainly a valid consideration in the weak 1900-1915 era, but I had Rusie 10th when he was inducted. I'll have to think about Big Ed some more.

Right, OCF. Alexander was my outside parameter.

Actually, throw in Rube Foster as a "definite", and recognize that Joe McGinnity is probably high enough now to work his way in eventually, and it is possible that a "20% pitcher" voter will determine that that's enough.

My view is that 19th century pitcher was at least as important that the early 20th century model (slightly less impact per inning, but more innings works out to more value).

Compare, say, Mickey Welch (most wins of the remaining 19th century pitchers) and Vic Willis (most wins of the 20th century pitchers on my list above). Welch pitched 20% more innings with similar K and BB rates and a similar ERA+. By what standard could I look at two, and conclude that Willis had more effect on his team's wins than Welch did?

I see nothing contradictory with concluding that 19th century pitching was less important, but there were simply more dominant pitchers. I put in Welch, McCormick, and Caruthers, definitely. I'd also put in McGinnity and Brown. Willis and Walsh are borderline, and beyond them, I haven't studied the pitchers enough to decide.

Pud Galvin is a Not Quite?

Luckily enough voters didn't agree with you. Galvin is much more qualified than guys like McGinnity and Griffith. I would put him as more qualified than Spalding and Rusie, personally (though they both belong as well)

Rusty, Jim gave us his methodology (sort of).

> Pud Galvin is a Not Quite? Luckily enough voters didn't agree with you. Galvin is much more qualified than guys like McGinnity and

Bob, you and your supporters always neglect that you posted your numbers in an inferior league and you were never among the top pitchers of your time in IP/season, despite having a short career. For one game, you would be great, but you just didn't play enough of them, either in a particular season, or overall, and when you did play you weren't (usually) playing against the best.

Joe, what do you do with Caruthers' part-career in the outfield? He didn't lead the league in IP, but he was nevertheless _playing_ more than other pitchers at that time.

Joe is probably not the best person to debate with, since he doesn't have a single pitcher in his Top 15.

Nonetheless, Joe wrote:

"you were never among the top pitchers of your time in IP/season, despite having a short career"

That gets it completely backwards, though. Imagine Player A, who gets 200 hits in 600 at bats and player B, who gets 300 hits in 400 at bats. Do we dock Player B for having fewer at bats? No, we praise him for doing more in less time.

Bob Caruthers pitched fewer innings than some of his contemporaries, but despite that fact accomplished more.

I think a good comparison this week would be Bob Caruthers (my pick as best pitcher available) and Joe McGinnity (the pitcher with the higest vote total).

WARP1/2/3

Bob: 98.5/69.8/73.4

Jim:

Although I haven't looked at ERA+, I like your pitcher categorization based on an eye-ball review. I would probably bump Joss to "not quite" b/c of his death-shortened career and move Welch up to "borderline." I also think that if McGinnity is "qualified," so is Griffith. But otherwise, the list looks pretty good to me.

I also agree with quite a bit of Jim's list, but let me comment on a few candidates.

1) Ed Walsh. The fact that Ed doesn't appear on Matt's list of good candidates suggests that he won't be a no-brainer for this group. I don't have an opinion on Ed yet, but those who support Ed should plan on his being a subject of debate.

2) Rube Waddell. Only member of Jim's "well-qualified" group that I'm certain doesn't belong there. His ERA+ and K's and other stats look great, but they are _never_ reflected as strongly in his wins as one would expect, and over the course of his whole career I can't believe that's coincidence. Where McGinnity and Griffith are 37.9 and 41.6 wins better over the course of their careers than their teams would have been with an average defense, Waddell is only 21.1 wins better, and he played on teams whose defense, over the course of his career, were _very slightly_ above average. That's a big difference. His stuff was surely fabulous, but it looks to me like he used it without much strategy, so that a pitcher with less stuff but much more guile, Clark Griffith, achieved much greater success when it came to getting victories for his team.

I agree with Jeff M. on Joss, McGinnity and Griffith.

I'd like to know how Jim sees Jack Powell and Chief Bender, who don't appear in his list. I don't have any opinion about Powell except that I need to form one (he's one of the guys whom this project has brought to my attention). From what I know of Bender, I think he's probably borderline at best, but I haven't looked at his numbers recently.

Just so we're not overlooking any pitcher candidates, this seems like a good time to rerun this list:

As a study aide for the pitchers, I put together a list for the first quarter century of the 60?6? pitching distance. Each of these 26 also topped 2700 IP. ERA+ is for their entire career; most of these pitchers had their entire career in the time frame.

At first glance, I like it, Jim.

Jim, I too like the list. But what's up with Bob Caruthers at 2.21. Is that a typo?

I would guess the 2.21 for Caruthers is about right -- he had only 1 bad season (1892, probably injury, I would guess, as it's before the pitching mound change yet he was only 28 -- does anybody know?), which dragged down his career ERA+ to 123 -- without it, it's above 130. If you like Koufax, you'll like Caruthers. If the number's right and not the positioning (which looks absurd) then he was the 11th best pitcher before 1925, which should put him into the HOM even without his Stovey-level hitting. It's a short career but a VERY good one.

I would guess the 2.21 for Caruthers is about right -- he had only 1 bad season (1892, probably injury, I would guess, as it's before the pitching mound change yet he was only 28 -- does anybody know?), which dragged down his career ERA+ to 123 -- without it, it's above 130. If you like Koufax, you'll like Caruthers. If the number's right and not the positioning (which looks absurd) then he was the 11th best pitcher before 1925, which should put him into the HOM even without his Stovey-level hitting. It's a short career but a VERY good one.

I still don't see it,

Trust your instincts, Joe. Trust your instincts. :-)

Over on the Hall of Fame thread on Clutch hits, this appeared:

<i>Posted 9:25 p.m., December 3, 2003 (#56) - Indian Bob Johnson

Interesting coupling of Indian Bob and Apollo, considering Apollo was widely quoted showing himself off as a racist b*****d.

Mr. Big Shot - I take it the job offer was meant for Bob, not me. I don't do "edgy" myself.

Posted 9:05 p.m., December 2, 2003 (#23) - Chris Cobb

For each team-season t, let Rt be the number of games played divided by the average number of games started for the two pitchers with most GS. With the right definitions, that is

I suppose in a few years we'll debate whether the conditions were tough (and Dave Stieb really was an all-time great) or whether it's just the luck of the draw.

I think closer to the former.

Yes, pitchers will always be the tougher choices. Were there really no great pitchers in the 1920s (other than an old Walter Johnson and a young Lefty Grove, both with their peaks in other decades)? Are there really 10-12 HoMers in the 1960s and '70s and only 2 in the 1980s and '90s? Or were those things driven by conditions? Considering our voting patterns so far, I would guess that Stieb, Key, Saberhagen et al are going to be on the outside looking in.

Dave Steib raw numbers:

WARP 3 - 89.2

There was a discussion here about Wins Above Team. Looking at the leaders in the last edition of Total Baseball shows this:

Pre-1893 Pitchers leading in WAT

Another early morning edit. Of course Dave Stieb's name is spelled i before e, since he has no "c" in his name.

BTW, many people (including probably me) have been misspelling Caruthers as "Carruthers." Let's stop that. He is a viable candidate and we should all take the time to spell his name right.

I've just looked up Stieb, and been convinced by him that WARP-whichever is another new-fangled nonsense. Raw wins and ERA vary across eras, but WPCT doesn't (except second-order, due to standrad deviation changes), and Stieb's was only .562, for a team that won 2 World Championships and was perpetually towards the top for almost all his career. 176 wins and .562 WPCT will not get him anywhere near my HOM, in any era, unless he was pitching for the 1899 Spyders.

I have a question (which shows my limited grasp of sabermetrics): is there a fairly straightforward method for establishing the average run value of a hit, such that one can estimate runs saved from hits saved?

I can see clearly enough that such an estimate would have to be rough, given that one doesn't have information (without play-by-play data) of the kind of hit prevented, or the context in which an out is made, but I'd still find it useful to have a rough-and-ready average value, specific to say, a single league in a single season.

Do I have any hope of there being a straightforward way to arrive at a decently reliable average value?

Stieb above Warren Spahn? Back to the drawing board, I fear.

Stieb above Warren Spahn? Back to the drawing board, I fear.

Silly to dismiss a metric just because its ranking of one player seems anomalous.

With all due respect, Jim's metric is, in general, a great deal more reliable than winning percentage.

Stieb's frequently a pitcher that stats-derived measures of quality rank a good deal higher than his wins, so it's no real surprise that Jim's does, too. It's Stieb, more than the system, that needs explaining.

All that said, no one metric is going to tell the whole story of a pitcher's relative value: this one gives a good starting place for thinking about how to balance the value of pitchers who are brilliant pitching in a fairly small number of innings with the value of pitchers who are workhorses.

The object of pitching is to win, so why is winning percentage unreliable? Particularly in the period before 1970 or so, when a high percentage of wins were complete or almost complete games. Over a season, random fluctuations can of course overwhelm it, but over a caeer, taking into account the strngth of teams and oppositions, it seems to me to measure the most important function of a pitcher, or any ballplayer -- winning.

The object of pitching is to win, so why is winning percentage unreliable?

Because a pitcher has, at best, only 50% of the responsibility (a little bit more than that during the early years of baseball).

Since my immediate interest is "short-to-medium career pitchers whose careers peaked between 1895 and 1915", as compared to still-eligible pitchers from the 1880's, I offer the following reduction of Jim's post #52:

Birth Year First Last Overall

The object of pitching is to win, so why is winning percentage unreliable? . . . Over a season, random fluctuations can of course overwhelm it, but over a career, taking into account the strength of teams and oppositions, it seems to me to measure the most important function of a pitcher, or any ballplayer -- winning.

_If_ one takes account of the things you mention -- the level of offensive and defensive support the pitcher receives -- I'd agree that winning percentage is a very important and useful metric. It's the one I'm studying myself, right now. But a lot of conversation on the threads I've been seeing about winning percentage lately has not been taking acount of those things. _Without_ those things being accounted for, I'd take Jim's metric over an unvarnished winning percentage number.

Above you mention that Waddell's defenses were above average...where are you getting that data? That would be very interesting to refer to.

I hope I said, and I meant to say, that Waddell's defenses were _about_ average. I got that data by calculating the league defensive efficiency for each year Waddell pitched, calculating his teams' defensive efficiency, park-adjusting it, and comparing the numbers season by season and calculating a career league DE and a career team DE for Waddell. DE isn't a complete measure of team fielding value, but it's close, and Waddell's teams generally look about average on FP as well. Here are the DE numbers for Waddell's career, as I have them:

Season -- Team DE -- League DE -- Team/League

Deconstructing Bob Caruthers

As we stumble along to the dreaded ?Candidate Gap? of the late 20?s-early 30?s ballots, Bob Caruthers is one of those Gray Area guys we have to get right. After a renaissance in 1916 balloting, I thought I?d better start trying to seriously assess him.

His is a unique career, making getting a handle on him difficult. He wasn?t like Ward, who pitched less and had a long career in the field as a gold-glover/base runner/brain. He wasn?t like Ruth, who pitched less and had a long career as an uber-hitter. He was simultaneously a fine pitcher and a fine hitter, with a very brief career after his arm went.

The easiest way to get a handle on him is to break him apart. Look at him as if his hitting career followed his pitching career, rather than being simultaneous.

How long did he pitch? He debuted in September 1884 and pitched until his arm went dead some time in 1892. About 7.5 seasons, a shorter career than any pitcher we?ve elected. His career ERA+ was 123, very good but far from our leading candidates. A similar career length and ERA+ to guys like Silver King and John Tudor. He was not a workhorse, never coming close to the league lead in IP.

How long was his hitting career? Taking each of his ten seasons and subtracting his games pitched from his total games and dividing by team games, shows him with only 2.6 full seasons off the mound. His career OPS+ is 135, again, very good but far from our leading candidates; not quite among the top 100 all-time. His defensive value (playing below-average RF 71% of the time) seems to be negligible. His base running seems average at best.

He had a nice peak, a serious MVP contender a couple times. I think Jennings had a better peak in a career of similar length. In his prime 1885-91 he played in 56% of his team?s games (34% pitching, 22% field).

A ten-year career at those levels isn?t quite up to HoMer quality. It all comes back to his uniqueness and his peak value 1886-87. He was a great hitter for a pitcher, but at the time this was hardly unheard of. I like guys who play. I don?t think there?s enough play here to put him over the top.

Dan,

Looking at Caruthers' value as a pitcher in terms of ERA+ underrates him quite a bit, because he was bringing that 135 OPS+ with him when he took the mound. That's a very good OPS+ for an outfielder, but it's totally stratospheric for a pitcher, even in the 1880s. To say, "He was a great hitter for a pitcher, but at that time this was hardly unheard of" is to vastly understate the difference between Caruthers and his contemporaries. Charley Radbourne has been talked about as a pretty good hitter, for a pitcher. His career OPS+ was 72. Pud Galvin has been talked about as a poor hitting pitcher. His career OPS+ was 46. So Caruthers is boosting the offense from the pitcher's spot by, say, 50-100 points of OPS+ when he takes the mound. The only player we've seen who created value for his teams in a manner similar to Caruthers was Al Spalding, who was both an exceptional pitcher and an exceptional hitter.

Yes, short career is a problem. No, he wasn't quite as dominant on the pitching mound as some. But because he hit like a good corner outfielder while he was pitching, his value to his team was highly similar to that of the very best moundsmen we have on the ballot whenever he pitched.

I'm not certain yet if he is the best pitcher available, but he's undoubtedly among the top 3.

Exactly right, Chris.

DanG seems to have split Bob Caruthers into two halves of the oreo, but the yummy creme seems to have slipped out, not adhering to either half.

7.5 years as a pitcher is one thing -- but a few others have pitched better, 2.6 years as a non-pitching hitter when he wasn't pitching is another thing -- but again just a plain chocolate wafer.

But where are all the at bats he had while he was pitching? This is what makes him a HoMer, and is what DanG is completely ignoring. The hitting AS A PITCHER is why we're buying the cookie.

No, he wasn't quite as dominant on the pitching mound as some. But because he hit like a good corner outfielder while he was pitching, his value to his team was highly similar to that of the very best moundsmen we have on the ballot whenever he pitched.

This may have already been done somewhere here, but has this statement been demonstrated to be true? And if true, for how long was it true?

This may have already been done somewhere here, but has this statement been demonstrated to be true? And if true, for how long was it true?

I don't know what exactly you would take as a sufficient demonstration, but see my post #25 on the 1916 ballot discussion thread.

Essentially, if you take Griffith, McGinnity, and Caruthers and compare their won-lost records to the won-lost record that a league -average defense would achieve, given the same level of team offense, Griffith comes out at 41.5 wins above average for his career, McGinnity at 37.5, and Caruthers at 40.5, in the fewest innings of the three. He had the best fielding support of the three, and we still have to deal with matters of AA discounts, but his effectiveness while pitching, when compared to average, is about the same as Griffith's or McGinnity's and is much higher than Waddell's or Willis's (who are at 21 and 25.5 for their careers, respectively).

There's more about the methodology in post 25 on the other thread and in some of the early posts on this thread. If you see problems with this approach, I'd be glad to hear about them.

Considering the following 22 pitchers (8 in the HOM and 14 not): Spalding, Ward, Galvin, Radbourn, Keefe, Clarkson, Rusie, Nichols, Mathews, Bond, McCormick, Whitney, Welch, Mullane, Caruthers, Hutchison, Griffith, McGinnity, Waddell, Willis, Joss, Young

These comparisons are using the BP numbers from the "Advanced Pitching Stats" section. They are more for fun than for serious decision making, though they may play a part when the decision-making gets very close. The numbers for best and worst are listed to give an idea of the range.

Best Defensive Support (DERA-NRA):

Chris:

I did a close reading of your post #25 and it is great work, a useful contribution, as always. However, as you pointed out, until you can factor in pitcher?s individual offensive support received, its value will be limited.

Also, I know you realize this, but Caruthers was the only pre-1893 pitcher on the list. While it seems the method should work for him, it?s possible there may be some adjustment needed in comparing him to pitchers of later eras.

We would expect Caruthers? run support to be a bit better than his teammates, because of his own bat in the lineup. I suppose there would be some way to quantify how many extra runs/wins he added to his record with his hitting over an average hitting pitcher.

Related to that, I think what I?m looking for is some way to transfer his amount of OPS+ above an average pitcher, over into his ERA+ of 123. How much would this boost his ERA+? Would he get into Joss/Walsh territory (142/145)? With his 2.6 additional years as a hitter he would be a shoo-in for the HoM. Or would he merely advance to Rusie/Waddell range (130/134)? That would probably leave him short of the HoM.

Having a preference for long careers, I have never voted for Caruthers. But if that great peak survives scrutiny, it?ll be hard to justify why Rusie and Spalding and Radbourn are in and not Bob.

Perhaps, too, it?s similar to what Bill James wrote about Vern Stephens; the raw numbers are so gaudy that they invite suspicion. Sure, Parisian Bob benefited from his circumstances, just as Stephens did or Chuck Klein did, or Larry Walker did. But after stripping away the excess, there could still be a great player under there. I?m honestly trying to see if that?s the case with Bob.

DanG,

I will be doing similar studies for the other pre-1893 pitchers as soon as I have time. I've started on McCormick. I will be re-doing post-1900 pitchers using Chris J.'s data (provided in the 1916 ballot thread) about actual run support, which should greatly improve the reliability of the results for those pitchers. If we had that kind of data for pre-1901 pitchers, we could quantify Caruthers' superior run support more easily, but we don't. Someday I guess Retrosheet will reconstruct the 19th century game for us, too. There may be a need to interpret this number differently from pre-1890 pitchers, but I don't have enough examples yet to have an opinion about that. More study is needed!

One way to incorporate Caruthers' offense into his ERA+ might be to use runs created.

From a 20th century perspective, Bob Caruthers is unique among pitchers for his OPS+ and his games played at other positions. I suspect that his type did not disappear. Perhaps Wes Ferrell in the 1930s and Bob Lemon in the 1950s are latterday examples. Ferrell hit OPS+ 100 and played 174 games when he did not pitch, mainly as pinch hitter; Lemon hit OPS+ 82 and played 155 games when he did not pitch, mainly as PH. If I am right, this is a general reason why it is important here to get it right with Caruthers.

In the 1916 thread, December 9, 2003 (#61)

Perhaps Wes Ferrell in the 1930s and Bob Lemon in the 1950s are latterday examples. Ferrell hit OPS+ 100 and played 174 games when he did not pitch, mainly as pinch hitter; Lemon hit OPS+ 82 and played 155 games when he did not pitch, mainly as PH.

Some other prominent post-1920 examples who come to mind are Red Ruffing (258 games not pitching, 81 OPS+), Bucky Walters (287 games not pitching, 69 OPS+), and Red Lucas (511 games not pitching, 84 OPS+). In fact, Lucas played most of his career games as a pinch hitter.

Someone noted that Babe Ruth will arrive here someday.

That was me, post #64 - 3 spots below Rick A. on the same thread. One doesn't actually have to use the name of G.H. Ruth for the identity to be clear. I was also making the point that changing from pitcher to everyday player may have been a sharply different experience after 1893 as opposed to before 1893, partly because it takes so much more physical effort to pitch from 60.5 feet.

By linear weights, the run value of turning an out into a single is about .75 runs. Of course, if you don't know if the hit is a 1B or 2B or 3B, it might be more, and it depends on whether we're talking 1905 where there are lots of 2B and 3B, or 2000, when triples have gone on life support.

Runs Created would give about the same answer.

Tom, thanks for this info!

Now a further question. I'm not conversant with linear weights, but I am conversant with runs created. I have used the basic formula (H+BB)(TB)/PA to calculate the run value of a hit (using average TB for a non-HR hit as the modifier for the TB figure), and the numbers that I get are notably below the .75 figure you provided from linear weights. For example, in the NL in 1885, 1 out changed to a hit of 1.271 bases was shown to have a value of .65 runs. In the AA of 1891, a hit was shown to have a value of .62 runes. In 1894, the hit was shown to have a value of .74 runs.

The basic RC formula significantly underestimates scoring in the early game, so I am uncertain whether these seasonal figures from RC or the .75 general figure from linear weights is likely to be more accurate. Does the RC formula accurately show that much of the scoring that was going on is not derived from the value of hits and walks, or does the RC formula underestimate the value of hits and walks because it misses their interaction with errors, passed balls, etc.?

Thus, would my analysis of the value of hits saved by fielders be more accurate if I used the RC values calculated for each season, or the general linear weights value? This may not be a question answerable in statistical terms, but I want to use the measure that seems most likely to be accurate. The results of RC are generally close enough to .75 that the difference in runs saved by a good defense that would result from using the different numbers will generally be less than 10, but I'd rather be as accurate as possible, of course.

Tom,

I see enough of the idea to understand the numbers I'm getting better. My numbers are lower across the board because of the prorating factor (I think I can work through that, or at least try), and the range of values I'm getting is a product of the variance in TB/H .

If you, or anyone else with know-how, has time to explain how exactly you prorate the end result, that'd save me the trouble of thinking it through myself, but I think I know enough to take the next step in converting hits saved to runs saved.

Upon further reflection, I am thinking that for what I'm doing, I don't need to deal with prorating, since for the calculations I am doing on the team level (steps 1 & 2 below) the number of outs is constant.

Here's what I have in mind to do.

1) start with team outs on BIP. Divide by team defensive efficiency, park-adjusted, to find team non-HR hits on BIP, park-adjusted. Call this A.

To take this out of stats and (maybe) into common sense:

It seems like we look at Caruthers and say, "Sure, he was good, but look at all the run support" - when in fact the extra run support is created by HIM! (he hit a ton, and the 'extra' OF for his team pummeled the puny P hitting of the other team).

Going over the top 5 P vote-getters, plus three key new arrivals:

McCormick 1878-87 265-214 118 ERA+ 2nd most wins, but 21-3 in UA

I'll bite. Just among those 8? No, let me add a couple other near-contemporaries for perspective. This does not match my recent ballots as I am trying to understand "new WARP" and I am beginning to explore what a ballot might look like if I gave up on McCormick and Bond. I may go back to my previous ballots until I "get" this, but just for today:

1. Ed Walsh

As I try to put together my ballot (and I'm _not_ going to be voting early this week!) I am turning over this question in my mind for the pitchers:

how important is exceptional durability in this period?

A lot of the discussion on this thread has focused on rate stats, esp. ERA+ and various win-based metrics. Jim Spenser's system takes account of IP, but it seems to be less important than ERA+. On the other thread, Redsox has expressed doubts about WS for pitchers because Joss isn't rated that highly, despite his great ERA+.

It seems to me that what WS, and also WARP, which rates Vic Willis pretty well, are telling us is that IP are more important than some (many?) of us believe. So I throw out the question. How important are IP in this period?

IP are important in any period. It's why I listed CG% in #24 above.

IMO, it's why Jack Morris is a viable Hall candidate even given his ERA. Sparky would leave Jack in the game sometimes when he was getting pummeled, just to rest the relief staff. With Morris having the ability to handle this extra load, it helped the other pitchers on the team perform better. Current rating systems give Jack no credit for this, perhaps because it's still speculative.

We all know bulk pitchers have value. (Nearly all teams would welcome a guy into the rotation who can pitch 200 IP with a 100 ERA+.) If Morris, or McGinnity, can go the extra mile when they're tired, even at a performance level below their normal output, they also save the team from having to give more innings to someone with a 85 ERA+.

I like guys who play. The more I think about him, the Iron Man is my kind of player. I don't think win shares is being unfair to Joss.

I guess I'm not certain, Dan, that there was the same need to _rest_ a staff on the shoulders of a bulk pitcher in this period as there obviously is today. Usage patterns vary so much from team to team that I'm not sure that these patterns are determined by the durability of the available pitchers or by the manager's style/philosophy.

Paul Wendt's posts have pointed out that pitchers were not being used in a regular _rotation_. Lack of a rotations suggests that there wasn't a shared understanding that a pitcher _needed_ so many days of rest between starts to keep his effectiveness and avoid injury. So what was it governed usage patterns?

Did Addie Joss throw fewer innings a season than Vic Willis because he wasn't as durable, or did he throw fewer innings because he was being managed differently? Did a lower workload contribute to his better performance when he did pitch? I don't feel that the answer to these questions is clear, and I'm looking for historical perspective.

Chris, does that matter? Value is value. A pitcher has no value while sitting on the bench. Doesn't matter why...?

To suggest why it _might_ matter -- I'm only raising possiblities, not arguing a position I'm sure should be adopted, let me present two pitching staffs from 1905: the Chicago Cubs and the Boston Beaneaters. For speed, I'm going to use ERA, not ERA+, but note that Chicago had a great defense and in a slight pitcher's park, Boston had an indifferent defense in a hitter's park.

Boston's pitchers

Oops. That previous post is FROM Marc... addressed TO Chris. Doh!

Howie Menckel (#82)

Much of the debate on the pitchers this week has cited W/L records and ERA+. I know there isn't much else to work with (other than Win Shares or WARP), but I'd like to work through here some cautions on those statistics.

We all know that W/L records have problems. Baseball's a team game, and while the pitcher may be the most important guy out there in any given game, he's not even half the game. (Bill James' estimate is about 1/3rd, and if you accept Win Shares, you also realize that the further back we go in MLB history, the smaller that fraction becomes.)

We don't judge position players on their cumulative W/L records, though looking at them may be interesting. I calculated some estimates based on prorating the team record onto the number of games played by each HOM player. Not surprisingly, the leaders in pct are the Boston 4 (Spalding 297-107 .735 is tops), with Wright (.715), Barnes (.689), and McVey (.685) following closely behind. Anson has the most "Wins" (1448-1028 .585) followed by Bill Dahlen (1319-1080 .550). Four of the bottom five position players are Davis (.529), Connor (.527), Delahanty (.521), and Hines (.493). "Those guys just didn't know how to win", but we elected them on the first ballot anyway. Worst was Jack Glasscock (735-969 .431); we elected him despite that because A) nobody pointed out this fact and we would have changed our minds had we known it, or B) we know that one player can't consistently carry a bad team to victory. (My point of course: neither can the pitcher.)

ERA+ takes Runs Allowed by a pitcher and corrects it for fielding errors (in a baroque fashion) and park effects. Up until a few years ago (before the DIPS revolution) it was accepted as a pretty good way to evaluate pitchers. It has a significant flaw though; it is subject to fielding effects. We know that fielders aren't evaluated by errors alone; that how many balls they get to is also important. However the extra balls fielded by good fielders but not by bad fielders do not show up as "penalties" on the bad fielders, but as hits and "earned runs" allowed by the victimized pitcher. ERA+ overrates pitchers with good fielders behind them and underrates pitchers saddled with bad fielders. I'm sure there are notable exceptions, but in general, good fielding is more likely to be found on good teams than on bad teams.

The seeming paradox of ERA+ is that given two pitchers with a similar career ERA+, the better pitcher is probably the one with the inferior W/L record, not because "he just didn't know how to win", but because he probably played on inferior teams and had to overcome the effects from inferior fielders to post that ERA+.

Paul in No. 92:

I don't count McCormick's 1884 because he basically was playing in the minor leagues, given the paucity of "major league quality" players. I don't see it as any better than what Pud Galvin was facing in the late 1870s, and if that had counted, Pud would have been voted into the HOM a decade earlier.

I ardently disagree with jimd's view that pitchers' W/L records are "negative information" (my words). Jim opines that of two pitchers with similar ERA+ but noticeably different W/L records, the pitcher with the worse W-L record is apt to have been the better pitcher.

I humbly ask Jim and others interested in the topic to look at the research I did related to a new stat I called Win Values. The articles are posted in the Authors section of the Primer website.

There is a definite finding that a pitcher's true value (in the sense I defined it) is a mixture of his ERA+ and his W-L record. And both pieces contribute in the same direction. A pitcher's value is higher if his ERA+ is higher and if his W-L record is better (not worse).

I understand the point that Jim was trying to make, but I believe he went too far in dismissing pitcher won-loss records.

I think that there is a temptation to overintellectualize some baseball statistics and lose sight of the the rather simple truth that the ultimate object is to win.

As long as everyone realizes that the pitcher has (at best) only 50% of the responsibility.

ERA+ and DERA

Prompted by jimd's reminders on the relative unreliability of ERA+, I decided to compare it to WARP's defense-adjusted run average (DERA), which is WARP's view of the pitcher's rate of run prevention. DERA is calculated by taking the pitcher's NRA (normalized runs allowed, which is a run average adjusted for park and season, with average set to 4.50) and removing the fielders' contribution, with the average again set to 4.50. WARP points out that if a pitcher's NRA is lower than the pitcher's DRA, it's safe to assume they pitched in front of an above-average defense. Here I've listed the HoM pitchers, current candidates, and soon-to-be-arriving candidates, in DERA rank order, with NRA in parentheses.

Ordered by DERA

Chris, the thing I've never understood about DERA is

> and removing the fielders' contribution

How do you know what the fielders' contribution is to remove? You go on to say

>if a pitcher's NRA is lower than the pitcher's DRA, it's

Chris Cobb:

Chris, the thing I've never understood about DERA is

> and removing the fielders' contribution

How do you know what the fielders' contribution is to remove?

Well, I don't know exactly what WARP does, but I infer that they do two or three basic things, both of which are predicated upon the fact that for pitchers, they calculate value in relation to league average first, and then estimate replacement level from there.

1) Find the team's defensive efficiency in relation to the league defensive efficiency to calculate hits saved and, by extension runs saved by the defense.

2) Examine the team's errors in relation to league errors and use this examination to further adjust the runs saved by the defensive efficiency calculation.

3) Adjust DERA based on the pitcher's contribution to or detraction from defensive efficiency (this step they may or may not take).

I infer this because I've been studying team defensive efficiency in relation to league defensive efficiency, and I reach the same conclusions that they do about which pitchers pitched in front of good or excellent defenses and which didn't. Also, I've calculated pitchers hits saved above or below team defenses, and my numbers match the delta H values WARP provides, so I conclude that their methods and mine are fairly close. I don't do anything with errors because I'm not sophisticated enough, but they are probably more sophisticated than I am, so I guess they are using that data in some fashion. I'm not sure about the third step because they have the data they would need to make such a calcuation (delta H), but the numbers are so small in relation to total defense in a career that they may not use those numbers, and I can't tell just from looking.

<i>You go on to say

>if a pitcher's NRA is lower than the pitcher's DRA, it's

Chris, I'm not questioning you or your methods, just hoping you can help me understand WARP, NRA and DERA. It seems to me that the next loop in the circular cause-and-effect that has been set up is what is called defensive efficiency. Particularly in one-man pitching rotations, how do we know that the defensive efficiency is not accounted for by some aspect of the pitching? Does WARP impose a DIPS analysis on an era that may or may not be DIPSy?

I doubt if I will ever understand this, but I appreciate your valiant effort to explain it!

Marc,

WHY I WON?T VOTE FOR BOB CARUTHERS FOR INDCUTION INTO THE HALL OF MERIT

Changing season lengths and changing pitching conditions make it hard to judge pitchers' durability in the 1871-1913 period. To give us another way of looking at durability that normalizes durability to some extent, I've calculated IP+, a stat that is meant to work for innings pitched sort of like ERA+ functions for ERA. The problem with calculating an IP+ is, of course, deciding what average IP is. We're interested in the durability of starting pitchers relative to other starting pitchers, so to calculate an average IP for each season, I've added up the innings pitched of all the "full-time" starting pitchers and divided by that number. The number of "full-time" starters per team has varied, of course. Here are the numbers I've used.

1876-1882 1 pitcher per team

I haven't done a study of 1871-1876 because it doesn't seem very important to assessing pitchers now on the ballot, and it would require more work to compensate for the wide variation in team games played.

I passed on it with a similar study of mine, too, because it's fairly easy to tell that it's not any different from the 1876-1882 era.

Good work, Chris!

Very interesting, Chris.

As I look over your numbers and the related stats, I see that it was an error for me to have Jim McCormick on my ballot, while excluding Vic Willis. Not sure exactly how I missed him before.

I don't know if I can justify 6 pitchers, but this definitely helps bump Willis up into mid-ballot territory.

Jack Powell was a workhorse with a career losing record (244-256) for undistinguished teams. His quick and dirty Pythagorean W-L record based on his RA+ is quite a bit better - I have it at 265-223. Is that enough to get him into the same conversation with Willis, McGinnity and Waddell? Chris Cobb or Chris J. - what do you have to say about Powell?

Re Powell:

I'd find it very helpful to see Chris J.'s run-support numbers for Powell.

I haven't done a comprehensive study yet, but in terms of durability he looks like a poor man's version of Vic Willis.

His career IP total is one of the highest of his era, but that's mostly due to career length. He pitched 16 years, but he did not reach league average innings in the second half of his career in any season. Leaving off his rookie campaign and his final three seasons, he earns an IP+ of 106 over 12 seasons, with only 7 seasons of above-average IP.

Compared to McGinnity's 131 IP+ over nine seasons (all nine above average) and Willis 123 IP+ over 12 seasons (11 above avg.), Powell's durability falls short, though he is outpaces Waddell (101), Joss (100) , and Griffith (97) on this measure.

DERA and ERA+ both rank Powell behind all five of the pitchers listed above. In fact, he's near the bottom of the group of all pitchers now receiving serious consideration for the HoM in both of these measures, so I expect that a look at his teams' defenses won't help him very much.

My take on Powell now is that he's worth talking about, but it appears to me that both McGinnity and Willis are better both in terms of durability and inning-for-inning value. I haven't weighed Powell's durability and long career against Waddell's and Joss's short-term brilliance yet, but since I have been rating both of them ahead of Willis, I don't see Powell getting close to my ballot. When he and Walsh bring the group of pitchers 1895-1915 receiving serious consideration up to seven, I think Powell will be number seven in the group.

I'd find it very helpful to see Chris J.'s run-support numbers for Powell.

No surprise - they stunk. From 1901-onward he started 369 games & had an RSI of 91.49. The only pitchers I know of with worse RSI & more starts are: Dazzy Vance (90.47), Bobo Newsom (89.84), Mark Langston 90.56), & Steve Rogers (88.99). He's just ahead Tom Candiotti (91.51). To be fair, I haven't RSI'd everyone over 369 starts, but I've gotten most of them (I figure I got 15-20 20th century pitchers to go). Year by year for JP:

1901..110

Right about W/L %. I changed how I figured this out. I've got one set of W/L numbers now (as opposed to two when I first brought it up). It's simpler & does a good job isolating run support alone from W/L records. It's pretty basic. First, pythag two different things: A) What a pitcher's W/L record would be based on his actual RA/9 IP if he had league average run support his entire career, and B) pythag using his actual RA/9IP & his actual run support. Figure the difference between the two pythags & apply that difference to his actual win loss totals. So a pitcher who won 15 more games with his actual run support pythag than his league average run support pythag would have 15 wins converted into losses in his career record.

Jack Powell, by this method, lost 15 wins due to his run support in real life. Convert those losses into wins & his 1901-12 record goes from 167-194 to 182-179. And his career record becomes 260-239 (leaving 19th century W/L records unadjusted for run support). That's what he is - a 260-239 pitcher with an ERA+ of 106. He's Frank Tanana. Here's the adjusted W/L records for other pitchers of this period (in all cases 19th century records are not adjusted, so with the exception of Addie Joss, they're all open to some notable problems):

Cy Young: 518-309 (+7)

Thanks for the comments, Chris & Chris. Now that I look at it harder, I don't see Powell making my ballot - but I'd still take him ahead of Leever, Phillippe, Chesbro, and Orth.

As for Chris J's little joke in the last sentence of #117: that adjustment is to the guy I referred to in another post as "Cy A. Young", whose career began in 1901. I'd have the "Cy A. Young" character (actual record 221-141, adjusts to better than that) ahead of Willis, Waddell, or McGinnity and about level with Walsh and Brown.

(Moved from 1917 ballot thread) Walsh basically pitched for seven years, didn't win 200 games, and is nine years from eligibility. Brown was dominant for six years on a great team - Joe McGinnity, anyone? Clark Griffith won more games on far crappier teams. Neither is a clearcut choice.

The new eligibles thread has Walsh eligible in 1920. Is that an "official ruling"?

Walsh and Brown are "clearcut" if you weigh ERA+ highly in your ratings. They are 4th and 12th (tied), respectively, among starting pitcher career leaders in adjusted ERA+ (> 1500 IP ).

This is not a question pertaining _exclusively_ to pitchers, but this seems the right place for it. Both Rob Wood and JoeDimino have expressed astonishment at Addie Joss's low level of support. Why should Joss's level of support be surprising given the level of support that Hughie Jennings, John McGraw, and Frank Chance have been receiving? All three of these players were truly great at their best, especially when one looks at their rate stats and their defense, but all had their period of greatness and their careers as a whole cut short by injury. Was Joss better than any of these three? Do rate stats carry more weight for pitchers than for position players when we evaluate? Should they?

Jennings, who as I see it has the most peak and the most career of these four, has been making my ballots, but the other three have not. Should I assess Joss differently from McGraw and Chance? Should I consider bringing all three onto my ballot? Cogent arguments for and against would be appreciated!

Chris, not to mention that several 19th century pitchers have been said to have had too short of careers for consideration, yet played just as long as Joss.

My personal opinion is that Joss' peak is not appreciably better than McGinnity's and NOT better than Walsh's or Brown's. Add to that the short career. Of course, many pitchers have short careers but still, I don't see Joss as particularly distinguished from several other pitchers. Tommy Bond and Jim McCormick are more easily distinguished from their contemporaries.

And as MattB pointed out on the catcher thread, if pitcher wear-and-tear makes a short pitching career more valuable than a short career at, say, SS or 3B, how about C? What is wrong with a guy who has a short career at C where the wear and tear is as bad as it is at pitcher?

I don't see Joss as particularly distinguished from several other pitchers.

I don't have him down as the best major league pitcher for any particular year he played, plus he had a very short career. I can't see him over Willis, McGinnity, Brown, Waddell or Walsh. He just wasn't durable enough.

Which non-pitchers have we seen with significant achievements as a pitcher, a la Babe Ruth?

Paul -