Ok, I finished Old Hoss. I still need to total Radbourn and Clarkson's performances against teams over/under .500 and do both measures for Galvin and I should do Griffith. Maybe this weekend. I still need to get my ballot together.

Stoopid BAR exam taking me away from the HoM...

Pre 93 HoM pithcers vs. other HoM pitchers and Welch and then Welch vs. HoM pitchers:

Radbourn         Clarkson        Keefe
Clarkson  9-8    Radbourn  8-9   Clarkson  9-10
Galvin   16-14   Galvin    7-5   Radbourn 14-10
Keefe    10-14   Keefe    10-9   Galvin    8-6
Ward      3-1    Rusie     3-4   Ward      3-7
Rusie     2-0    Caruthers 2-1   Rusie     5-2
Caruthers 0-1    Young     1-1   Young     0-2
Young     1-0    Nichols   0-1   Nichols   3-3
Nichols   0-1
total    41-39   total    31-30  total    42-40

Welch    10-17   Welch   6/7-13

Welch
Galvin   26-11
Radbourn 17-10
Ward      4-7
Clarkson 13-6/7
Rusie     2-0
Nichols   0-1
Caruthers 0-2
total    62-37/38 *

*different total because it now includes record against Caruthers and found 1-1 record to add to Clarkson and question over how to count a forfeit when the game was tied.
Clarkson's total is the lowest for two reasons:
1. He played with Radbourn for several years, then Nichols and Young.
2. It looks like he was skipped against HoM pitchers and other top starters in some years, at least from the game logs.

Jim - I believe the bottom 15% of regulars in the majors are the definition of replacement level

I'll agree with that as an AVERAGE. That is, the bottom 15% of regulars should average to the replacement level, i.e. to 0. The bottom 15% of the regulars in 1895 using BRARP+FRAA average to -17 runs. That is an overcorrection, as Jeff M has indicated earlier. The original problem is about 10 runs the other way.

This overcorrection results in 27% of the regulars being below replacement level, not the expected 5-10% (half of 15%). Using unadjusted BRAR+FRAR, 5% of the regulars are below replacement level. The cure is worse than the disease. Also, an adjustment of approximately 10 runs places 7% below replacment level.

Total BRARP+FRAA for NL regulars in 1895. (Old pages.)

RF: Thompson +86, Keeler +66, Bannon +43, DuMiller +41, 
    Tiernan +37, Ryan +23, Donovan +12, Dowd +7,  
    Blake +1, Treadway -9, Gettinger -14, Hassamaer -25

CF: Lange +65, Stenzel +64, Griffin +58, Hamilton +54,
    VanHaltren +33, Brodie +27, Duffy +25, Hogrreiver +6, 
    Wright -9, Abbey -12, McAleer -14, Brown -21

LF: Delahanty +79, Burkett +73, Kelley +61, Selbach +39,
    Clarke +32, Cooley +31, ESmith +16, Hoy -1,
    Anderson -4, Wilmot -6, McCarthy -8, Burke -9

SS: Jennings +117, Dahlen +45, McKean +37, GSmith +31,
    Long +21, Fuller +19, Corcoran +18, Ely +6, 
    Sullivan +1, MCross -11, Schiebeck -20, Shugart -24

3B: McGraw +63, Davis +51, Joyce +40, LCross +34, 
    Nash +29, Everitt +19, Shindle +18, Collins +13, 
    Clingman +13, Latham +6, McGarr -10, DoMiller -11

2B: Crooks +49, McPhee +47, Lowe +37, Quinn +29,
    Childs +24, Hallman +24, Gleason +5, O'Brien +3, 
    Bierbauer 0, Stafford -8, Stewart -8, Daly -9

1B: Cartwright +65, Connor +44, Beckley +44, Anson +31,
    LaChance +25, Ewing +20, Tebeau +14, Doyle +14, 
    Tucker +3, Spies -5, Carey -22, Boyle -40

There are just too many players that are significantly below 0 that are also playing 100+ games (see prior chart on previous page).

Good example of why one shouldn't elect anybody but Ruth/Cobb/Wagner/Johnson on the first ballot. We should try to avoid post-election regrets in future, and even if it's a weak year make marginal candidates (Wheat?) run the gauntlet like Wallace, McGinnity and Sheckard.

We know a HUGE amount more about these guys than we did when Clarkson was elected. My view is, the late 80s were just a good period for pitching, with Welch one of the best, but strategy was obviously a big deal for these guys, and pitching to the score was one of the strategies they used. But everything we learn says there was no significant Merit difference between Welch, Clarkson and Keefe (Caruthers, just as good but not quite as much, is a separate question because of his hitting.)

Wouldn't purposely underrating new eligibles be an example of strategic voting?

pitching to the score was one of the strategies they used.

I don't see how "pitching to the score" could ever make sense as a strategy. Since no lead in baseball is safe, why wouldn't a pitcher always want to concede as few runs as possible?

I don't see how "pitching to the score" could ever make sense as a strategy. Since no lead in baseball is safe, why wouldn't a pitcher always want to concede as few runs as possible?

I agree. It makes no sense. Unless you're winning by a truckload of runs, I can't see a easing up at all (the latter point is relative of course: pitchers threw as hard as their respective eras dictated).

We should try to avoid post-election regrets in future, and even if it's a weak year make marginal candidates (Wheat?) run the gauntlet like Wallace, McGinnity and Sheckard.

I agree Wheat should run the gauntlet and won't be high on my ballot, but I don't think we should suppress his vote if we honestly feel he belongs at the top.

The decision making of this group, as with any group, works best with the Delphi method, whereby several rounds of discussion are used to reach an informed consensus.

Since we don't unelect the elected, and work in only one temporal direction, this method doesn't have time to work properly if we elect people on the first or even second ballot.

Hence there is a danger of marginal candidates, who we would not have elected on mature consideration, slipping through. Walsh, Flick, Magee and Gore, to me, are HOMers where a few years' thought might (but not necessarily would) have caused us to revise our collective view.

Probably none of those 4 is a serious mistake, but we may yet make one if we're not careful about this.

We know Matty did "pitch to the score" -- it's reasonable to suppose these 80s guys did too. Makes a lot more sense than you think if you're pitching every 2nd day and there are no relief pitchers.

Probably none of those 4 is a serious mistake, but we may yet make one if we're not careful about this.

No disagreement with you there.

We know Matty did "pitch to the score" -- it's reasonable to suppose these 80s guys did too. Makes a lot more sense than you think if you're pitching every 2nd day and there are no relief pitchers.

But pacing oneself is not really the same as pitching to the score. As long as you are not giving up a run, a hurler can afford not to throw his hardest stuff and give up a hit in a non-pressure situation. The object, however, is not to give up a run (unless there is enough of a lead to give you some breathing room).

Re: Pitching to the score.

As I have been digging through the gamelogs of 1880s pitchers, I now think that this occurred to some degree. This is all based on a "feeling." I think in games where the score got out of hand early, the pitcher on the winning side would relax and not worry as much about allowing a few runs. Remember, these teams would often score 10-15- or more runs in a game. That could give a large margin for error.

I posted this earlier, but did not receive a response, but what would people consider proof of pitching to the score?
Would it be a pitcher allowing 1-3 runs in the 7th inning or later with a lead of 5 runs or more?
What do people think?
Also, does anyone know what newspapers would be the best for finding inning-by-inning scores for the 1880s?
This is what I am looking for (example)

team  1  2  3  4  5  6  7  8  9  r  h  e
away  0  2  0  1  0  0  3  0  0  6 10  3
home  1  0  0  4  0  1  1  0  x  7 12  3

Does anyone know where I can find this info?

Does anyone know where I can find this info?

The Brooklyn Daily Eagle is on-line:

http://www.brooklynpubliclibrary.org/eagle/index.htm

You might want to first ask Paul Wendt or someone at retrosheet if they have information like this cataloged somewhere.

Wouldn't purposely underrating new eligibles be an example of strategic voting?

Depends on how its done.

Let me try to explain it this way. Now I'm sure we've all realized that it's often hard to tell the difference between who should be in 6th place & who should be in 7th place - heck sometimes its hard to figure out who should be in 4th and who should be 14th. Talent often gets bunched up but good.

Now, when someone comes onto the ballot, it's often hard to figure out a precise single slot that he belongs in. Often times a person's attitude (at least in my case) is more like: "Well, he's worse than that guy at #6, but better than that guy at #12 - between there it gets tougher."

Now, what I've often done in the past - & what I assume's being promoted here - is the idea that you put him in the proper clump, but don't put him any higher than you're sure he's earned. In the above hypothetical, you put him between 7-11, but don't put him at 7 or 8 unless you're absolutely damn sure he's better than all the other guys in the clump. If over time you become convinced of it, but him there.

It's not strategy in that there's no guarantee that the person will rise up - sometimes, upon further examination a player's stature decrees. Othertimes he gets overshadowed. Also, there's no guarantee that a player will start low on your ballot if you adopt this approach. For example, on my ballot, Sherry Magee debuted in the #2 slot behind Home Run Johnson because I was convinced he was better than the rest. If I feel the same way about Zach Wheat, then sure I'll put him that high.

It is strategy if someone pre-determines to move Zach Wheat or anyone else a few slots lower than he feels Wheat deserves. That's my feelings anyway.

Also, does anyone know what newspapers would be the best for finding inning-by-inning scores for the 1880s?

I've seen the Chicago Tribune on microfilm - they gave boxscores. Beware: Old newspapers often listed the home team on top and then the road team on bottom, despite the fact that the road team batted first.

Wouldn't purposely underrating new eligibles be an example of strategic voting?

Yes, although I think that rating a new player conservatively, given the range of values one sees as possible for that player is defensible. To put it another way, taking one's level of confidence in an assessment of a player's merit into account when ranking players is legitimate: better to put the player you are sure of higher. Rating a player below where you think he legitimately rates to keep him from getting elected "too soon" _is_ strategic voting and therefore unconsistitutional.

I don't think the "Keefe and Clarkson elected fast, Welch not" case offers justification for making new players run the gauntlet, for two reasons.

First, and most simply, I don't think there's anything to regret about our election of Keefe and Clarkson; they are deserving. What I regret is that we didn't elect Welch back then, also.

We didn't elect Welch then because the accounts of value that the electorate found persuasive showed Welch as significantly below Keefe, Clarkson, and Radbourn. Galvin also had trouble getting elected, but the first major change in the evaluation -- the replacement of ERA+ by WARP's DERA as the most widely accepted measure of pitcher excellence, brought him in, but it didn't help Welch's case. The electorate's view of him began to change later.

That leads to the second reason why the 1880s pitchers aren't evidence for a "run the gauntlet" approach. They are the hardest group of players we've had to deal with, and for no other group has so much new data and analysis been gradually provided for us as the project has gone forward. To review: When Keefe and Clarkson were elected, WARP was not widely used by the electorate. ERA+ and innings pitched were the main benchmarks (Karlmagnus, of course, has held fast to wins as a measure, and is being justified in Welch's case by the developing analysis). WARP became widely accepted a few elections later, helping Galvin's case. We didn't have Chris J.'s RSI data; when Chris J. brought his RSI to the project, he didn't have Retrosheet data to work from for 1880s pitchers. When he was able to provide RSI data for 1880s pitchers, we first had new data that gave strong support to Welch. When WARP significantly changed its assessment of 1880s pitchers several months after Galvin was elected, its assessments came under sustained scrutiny, leading to other attempts to measure the role of fielding support in pitchers' performances. That also has helped Welch's case. It's the Retrosheet data, now analyzed beautifully by Kelly in SD as well as Chris J., carefully targeted studies of defensive support, and renewed skepticism about WARP's assessment that have revived Welch's candidacy. A whole lot has happened: I've learned more about studying the game and analyzing its statistical record from dealing with this group of pitchers than I have from any other part of the project, and it's been very satisfying.

But to return to the point about making a candidate like Zack Wheat run the gauntlet: how likely is it that our views of 1910s outfielders will be drastically reshaped by the advent of new evidence and analysis as they have been for 1880s pitchers? Voters will disagree about his value, because he's not a shoo-in candidate. We'll differ about whether or not there are too many left-fielders elected, about defensive value, about whether or not WARP's competition adjustments are too harsh for to the National League (and whether or not they are different in the new WARP). But, given our various conclusions about these various debatable issues, I think we'll each be able to place Wheat with a fair degree of confidence in relation to other eligibles. If the electorate doesn't think he's worthy, that will hold him back from election, and then we'll have a number of years to sort out his position among the borderline candidates, but there's no value in obscuring where he stands by deliberately holding him back.

I will say, taking a quick glance at Zach Wheat's stats, that I can't imagine that I wouldn't put him in an elect-now slot in '33.

I will say, taking a quick glance at Zach Wheat's stats, that I can't imagine that I wouldn't put him in an elect-now slot in '33.

Not really an impact player, though he did have a long career of fine quality. Players were becoming a little more durable during his era compared to earlier generations so he might superficially stand out compared to earlier players in many voters' eyes.

As I have been digging through the gamelogs of 1880s pitchers, I now think that this occurred to some degree. This is all based on a "feeling." I think in games where the score got out of hand early, the pitcher on the winning side would relax and not worry as much about allowing a few runs. Remember, these teams would often score 10-15- or more runs in a game. That could give a large margin for error.

I 100% agree with this. If it happened that often, then a player such as Mickey Welch may have inflated his ERA in games such as this where there was a comfortable lead.

I posted this earlier, but did not receive a response, but what would people consider proof of pitching to the score?
Would it be a pitcher allowing 1-3 runs in the 7th inning or later with a lead of 5 runs or more?

Personally, I would think it would have to be at least a six run lead.

If he had only a four run lead, that would be a little too close for comfort.

And its Clark Griffith Time, yeeeeaaaaahhhhhh.

Clarkson  0-1
Young     1-6
Nichols   6-7
Rusie     1-3
McGinnity 2-0
Plank     2-3
totals   12-20

That's it. I noticed a distinct pattern in 1898. Either he had the luckiest rotation scedule or he was not being pitched against the top pitchers:
Young, Nichols, Rusie, and Willis pitched against Chicago 4 times each that year and Griffith faced Willis once. That's it.
I am going to scan his years from 94-03 again to see if the pattern continues.

Noooowwwww, It's Time for the Clark Griffith Shooooowwwww. Yeeeaaaahhhhh.

This is the first time I have done this for any player: I counted the number of times each season a HoM pitcher pitched against a Griffith team and in how many of those games did Griffith pitch. Also, I added 1-2 top pitchers from top teams to get to a total of 5 pitchers each year, except for 1901 AL where I only used 4.

Summary: I looked at 9 seasons where Griffith pitched the most: 1894-1902. I totalled the number of HoM pitchers who pitched against his teams and how often he pitched in those games. Also, to bring the total to 5 pitchers each year and to include top pitchers from a particular year, I added one or two others.
The additional pitchers were: 1894 - Stivetts, 1895 - Hoffer (31-6 for Bal) and Hawley (31-22 for Pit), 1896 - Hoffer (25-7), Killen (30-18 for Pit), and Dwyer (24-11), 1897 - Corbett (24-8 for Bal) and Breitenstein (23-12 for Cin), 1898 - Willis and McJames (27-15 Bal), 1899 - Willis and Hughes (28-6 Bro), 1900 - Philippe and Tannehill, 1901 - Miller (23-13 Det), and 1902 - Waddell, Joss, and Powell (21-17 StL). Other pitchers could have been chosen but these were good pitchers who pitched a lot which was what I was looking for.
Results: Over the 9 years, these pitchers pitched against his team 190 times and Griffith was on the mound for 45 of them or 23.6% of possible times. I don't know if this is average or above or below.
However, there are some interesting patterns in the data:
Young faced Griffith only 6 out of 38 times he pitched against ChiNL/AL.
Rusie - 3/15
Nichols - 13/28
McGinnity - 2/14
Plank - 1/8

Conclusion: Unless it was Nichols, Griffith didn't face HoM pitchers. In fact, he faced Young ONE time in five years - 97-01. Out of 304 starts in the 9 years, he faced a HoM only 25 times (27 if include his two starts against Clarkson in 1894). Either he was not being pitched against these pitchers on manager's choice OR every other HoM starter was being kept from facing Griffith by their managers.

Note: The rest of the starts needed to get the results shown in post 317 came in 1903-5.
Intuitively, this doesn't impress me. What do other people think?

Oh, I should have the W-L records for when Clarkson, Radbourn, and Griffith pitched against +/- .500 teams and by league position up on Saturday afternoon or Sunday morning.

Griffith faced a HoM 25 out of a possible 103 times and only 12 / 75 if Nichols is removed.

This is too time consuming to do by hand again and I don't have a sortable statistics program so I hope someone else has similar info so we can compare my findings.

The decision making of this group, as with any group, works best with the Delphi method, whereby several rounds of discussion are used to reach an informed consensus.

Since we don't unelect the elected, and work in only one temporal direction, this method doesn't have time to work properly if we elect people on the first or even second ballot.

If you think that an upcoming candidate could be a mistake, speak your mind- then we can have several rounds of discussion before they even become eligible for election. It looks like the discussion is already underway for Zach Wheat. As I consider career accomplishments as part of my rating system, Wheat will very likely be in the #2 spot on my ballot in 1933 (behind the no-brainer).

Over the 9 years, these pitchers pitched against his team 190 times and Griffith was on the mound for 45 of them or 23.6% of possible times. I don't know if this is average or above or below.

He started 304 of his team's 1263 games in those years, so that's means he should've started 46 of the 190 HoMer games. So he's pretty much on line, though I would've guessed it would've faced more HoMers than the math would say because good pitchers often matched up against each other.

In my mind, there's no way that Wheat belongs in the HoM before Heinie Groh does. Groh had a far greater peak and played third base when it was still more defense-oriented than it is now. Plus, the attrition level for third basemen was greater than for the outfielders, so they didn't (they still don't) compile the same amount of career stats.

But I assume people are going to see 380 Win Shares! and ignore all of that (though James feels Groh is more deserving, FWIW)...

BTW - great work Kelly in SD.

Griffith faced a HoM 25 out of a possible 103 times and only 12 / 75 if Nichols is removed.

Yea, but I don't see any reason to remove Nichols. His being matched up againt Nichols largely makes up for the way he ducked the others.

Moving on to another pitcher - here's more info on Mickey Welch.

The main dilemma of Welch is how could a guy with that ERA+ get that W/L record? Luck? Pitching in a pinch? I thought I'd check one thing: compare what percent of his runs were unearned versus that of the rest of the pitchers he played with to determine if he allowed more or fewer earned runs than allowed in his career. This is the same approach that figured that Rube Waddell allowed almost 50 more unearned runs that he should've based on his RA and the team's % of UER.

In the below list, a "-X" is bad for Welch - it means he had X fewer earned runs than should be expected. Conversely, "+X" is good for him. Here's his career:
1880 -8 (he pitched 77.8% of the innings that year)
1881 -2
1882 -3
1883 -7
1884 Even
1885 +8
1886 -5
1887 +17
1888 +8
1889 +5
1890 +18
1891 Even
1892 +3

Total: +34 or +42 depending on how you count 1880. The reason I'm equivicating on 1880 is the purpose of this is to get the pitcher's UER in line with team norms, and in 1880, Mickey Welch just frickin' WAS the team norm. His ERA moves from 2.71 to either 2.65 or 2.63, depending on how you slice 1880. His ERA+, based on my wildly rudimentary math, becomes 116 either way.

Wheat and Groh will both be on my ballot, but neither of them beats Beckley or Welch, if they're still around. I expect the top to look like:

1) Johnson
2) Beckley
3) Welch
4) Wheat
...
about 12) Groh

Clarkson's total is the lowest for two reasons:
2. It looks like he was skipped against HoM pitchers and other top starters in some years, at least from the game logs.

Do you see Clarkson & teammates pitching out of expected order?
At least for 1885-86, when Chicago was the top team and Clarkson the top pitcher, isn't it likely that other managers rearranged their orders against the White Stockings?

Patterns of usage for ace pitchers - it will be a challenge to interpret those clearly

Pitching to the score, in a pinch - another challenge

Either one, and there must be others in the HOM air, might alone be the subject of a SABR Convention presentation (theater or poster) or an article in "By the Numbers" (SABR Statistical Analysis Committee newsletter).

P.S. I don't know whether these subjects have been handled by internet authors who focus on sabermetric studies. I guess that I would have read about it here, where sabermetrics is applied, but that is only a guess.

In my mind, there's no way that Wheat belongs in the HoM before Heinie Groh does.

Well, if it makes you feel better, Groh should debut around the middle of my ballot. Great player at his best, and the attrition rate for 3B was very high at the time. Wheat's longevity and consistency, however, still rate him higher in my system, even though he was a LF.

Happy birthday Larry Doyle

Apropos of nothing...oh, wait, apropos of running Zack Wheat through the gauntlet sooner rather than too late... Here's how I would rate the players who MIGHT be eligible in 1933 (obviously, up to three of them might be elected by then).

1. Pearce
2. Jennings
3. Santop
4. Pike
5. Groh--IOW I'm with John
6. R. Foster
7. Mendez
8. Chipper, er, ah, Charley Jones
9. Bond--Tommy, not James
10. Doyle--happy birthday
11. Wheat
12. Browning
13. the black Hughie Jennings--Dobie Moore
14. Childs
15. Williamson

And if 3 of the above are indeed elected (rather than, say, Van Halen and Beckley) then Bill Monroe, Jim McCormick and Rube Waddell are poised to leap onto my ballot.

Wheat, IOW, is part of a large "klumpp" (thanks to Chris to a perfectly usable, utterly unambiguous term) of LFers who are somewhere below the NB (i.e. Delahanty, T. Williams, B. Bonds) level and the near-NB (Jesse Burkett, Al Simmons) level. Chronologically they are C. Jones, Stovey, Kelley, Sheckard, Magee and P. Hill. Jones and Magee are already in my PHoM and there is indeed a thin line between those two and the rest, including Wheat, and the exact location of the line will surely change. But it is hard to picture all of them as PHoMers. I think they will become PHoMers (or not) as follows:

Wheat--may wait 15-20 years or more
Stovey--may wait longer than that
Kelley--ditto
P. Hill--probably not
Sheckard--ditto

Again there is a very very faint line anywhere within the whole group and they are all very borderline. But Wheat falls at or near the top of the klumpp and will likely be a PHoMer someday.

Groh, OTOH, will be a PHoMer long before I get to Zack. Like I say, I'm with John. I see Heinie as the second best 3B in ML history at the time of his eligibility.

Groh, OTOH, will be a PHoMer long before I get to Zack. Like I say, I'm with John. I see Heinie as the second best 3B in ML history at the time of his eligibility.

I don't think the analytic systems do a good enough job with the pre-Mathews third basemen.

I have Groh as the third best third basemen at the time of his retirement, BTW (but that is certainly arguable).

I'm also with sunnyday2 concerning the "klumpp."

Well, if it makes you feel better, Groh should debut around the middle of my ballot.

I'll take what I can get, I guess. :-D

BTW, Groh may be my new "Pearce," so be afraid.

Be very afraid! :-0

On Wheat and Groh:

Assuming Pearce is elected this year and Santop next year, the top of my ballot in 1933 will probably read:

1) Walter Johnson
2) Zack Wheat
3) Heinie Groh

I think both of them are clear HoMers.

Someone wrote of Larry Doyle:

C+ defender by Win Shares, . . .
Regularly in the 2B defensive Win Shares leaders,

Strange. Does this indicate a Bill James blunder?

I've got this pecking order among outfielders, leaving Cobb, Speaker, and Negro Leaguers (e.g., Torriente) out of it:

Heilmann
Wheat
Goslin
Van Haltren
Ryan
Duffy
Cravath
Burns
Hooper
Youngs
Veach
Thomas

Some of those could be shuffled around a little - maybe Thomas ahead of Veach. Tiernan and Browning should really be in there somewhere, too.

As for Groh - 3B is a tough position to deal with. As an offensive player, I've got him about even with Evers, behind Childs, and well behind Doyle - but those are all 2B, not 3B. In my mind, Groh certainly has more bat than either Jimmy Collins or Pie Traynor, and I'd take him ahead of Leach as well. Frisch is an interesting comparison that I'm not ready to decide yet.

Strange. Does this indicate a Bill James blunder?

1907: Didn't make the top five
1908: Didn't make the top five
1909: #4
1910: Didn't make the top five
1911: Didn't make the top five
1912: #3
1913: #4
1914: #5
1915: #4
1916: #2
1917: #1
1918: Didn't make the top five
1919: #4
1920: #5


Paul, to answer your question, I don't think so. He was in the middle of the pack for NL second basemen for most of his career. Not terrible by any means, but nothing great.

OCF: What are your thoughts about Max Carey and Edd Roush?

Here's my list, including Carey and Roush, and some other guys:

Torriente
Heilmann
Wheat
Goslin
Carey
Roush
Van Haltren
Poles
Pike
Duffy
Ryan
Cravath
Burns
Hooper
C. Jones
F. Jones
Veach

Chris - I see Roush as fitting right into the middle of the Van Haltren/Ryan/Duffy cluster. Carey seems to be a pretty good match for Fielder Jones, at least on offense, which would be down around the bottom of my list. What would you say for his defense - truly exceptional or just another CF?

From the Baseball Prospectus Glossary:

BRARP [ Details ] [ Return To Top ]


Batting runs above a replacement at the same position. A replacement position player is one with an EQA equal to (230/260) times the average EqA for that position.

Link: http://www.baseballprospectus.com/glossary/index.php?context=3

But now I'm noticing the BRARP isn't listed anymore for hitters, what happened to it? It was there before right? Or was I hallucinating? I just used BRAR without even thinking . . .

So yeah, the numbers I spent 2 1/2 nights inputting aren't really of any use - I'm pretty damn pissed about that. Thanks for pointing it out guys.

But now I'm noticing the BRARP isn't listed anymore for hitters, what happened to it? It was there before right? Or was I hallucinating? I just used BRAR without even thinking . . .

No hallucinations, it was there before -- it's still there on some of the cards that haven't been updated yet. (see http://www.baseballprospectus.com/dt/mcgrajo01.shtml.) They've taken it out in the update, for whatever reason. I'm sorry you went to all that work for nothing.

I just hit that link and it's not there for McGraw either. What the hell were they thinking. Brutal.

Re: Max Carey and his defense:

Win shares sees him as a great defensive outfielder. His grade is A+, his total defensive win shares are third all time among all outfielders (behind Speaker and Mays).

With WARP, it's harder to tell, especially since they are in the midst of a system-change. Assuming defensive numbers don't change much, here's how Carey stacks up in FRAA in W1/W3 against other relevant CF candidates:

Speaker 156/124
F. Jones 99/37
Carey 81/47
Griffin 62/-30
Cobb 56/25
Thomas 31/8
Ryan 27/-55
Duffy 13/-68
Van Haltren 6/-81
Roush -6/-36

So I'd judge that WARP sees him as an exceptionally good defensive centerfielder -- below Speaker (like everyone), comparable to Griffin and F. Jones, and well ahead of the average defensive centerfielders.

He lands so high for me because he has a long career -- I think his career looks rather like what Fielder Jones's would look like if he'd played another three years as at least an average player.

WinShares perspective on the above outfielders, plus some others who were retired by around 1930:
players ranked by ws/1000 (min 12000 innings)

player   pos def inn   ws  ws/1000innings
Speaker*  cf  23880  117.8      4.93
Carey     cf  21683   94.8      4.37
F. Jones  cf  15700   67.4      4.29
S. Brodie cf  12401   51.1      4.12
H. Duffy  ##  14638   58.3      3.99
M Griffin cf  12854   48.0      3.74
Sheckard* lf  18160   66.3      3.65
Kelley*   lf  12537   44.7      3.57
Hamilton* cf  13814   49.0      3.55
Paskert   cf  14634   51.7      3.54
Roush     cf  16519   58.1      3.52
Thomas    cf  12709   44.5      3.50
Clarke*   lf  19334   64.1      3.32
Lewis     lf  12512   40.1      3.21
Milan     cf  17133   55.0      3.21
Cobb*     cf  25834   82.6      3.20
Ryan      xx  17175   54.1      3.15 
Hoy       cf  15505   48.9      3.15
Beaumont  cf  12498   39.0      3.12
Veach     lf  15343   47.0      3.06
*HoM or no-brainer
## 40Ï/34%lf/26%rf
xx 49Ï/20%lf/31%rf

These were all the 12000 innings outfielders with more than 3.00 WS/1000 who careers were over by 1930(or so).
Other outfielders-candidates follow:

Leach        cf  9613   44.6    4.64
Van Haltren  cf 15863   46.0    2.90
Burns        lf 16303   46.9    2.88
Wheat        lf 20589   58.1    2.82
C.Jones      lf  7771   20.7    2.67
Hooper       rf 20314   53.1    2.61
Goslin       lf 19105   49.3    2.58
Browning     **  8868   22.8    2.57
Tiernan      rf 12911   31.0    2.40
Cravath      rf  9270   19.0    2.05
Heilmann     rf 13846   24.8    1.79
** 49Ï/48%lf/3%rf

Current HoMers (w/ at least 6695 def OF innings):

Gore       cf  11571   50.4    4.35
Sheckard   lf  18160   66.3    3.65
Kelley     lf  12537   44.7    3.57
Hamilton   cf  13814   49.0    3.55
Hines      cf  11234   38.7    3.44
Clarke     lf  19334   64.1    3.32
Stovey     lf   8318   26.5    3.19
Delahanty  lf  11645   33.1    2.84
Magee      lf  16390   44.4    2.71
Keeler     rf  17851   47.8    2.68
O'Rourke   **  12111   32.3    2.66
Jackson    lf  11360   29.7    2.61
Flick      rf  12758   31.8    2.49
Crawford   rf  19934   47.2    2.37
Thompson   rf  12468   28.4    2.28
** 32Ï/53%lf/15%rf

GREAT BIG comparison between Clarkson/Keefe/Radbourn/Welch (I haven't forgotten about Galvin, but his career takes the most work and is the most different from the other four)
This will have career records against HoMers, by finishing position, and against +/-.500 teams. Records are from Retrosheet so they won't match exactly to career w/l.

Radbourn         Clarkson        Keefe
Clarkson  9-8    Radbourn  8-9   Clarkson  9-10
Galvin   16-14   Galvin    7-5   Radbourn 14-10
Keefe    10-14   Keefe    10-9   Galvin    8-6
Ward      3-1    Rusie     3-4   Ward      3-7
Rusie     2-0    Caruthers 2-1   Rusie     5-2
Caruthers 0-1    Young     1-1   Young     0-2
Young     1-0    Nichols   0-1   Nichols   3-3
Nichols   0-1
total    41-39   total    31-30  total    42-40

Welch    10-17   Welch   6/7-13

Welch
Galvin   26-11
Radbourn 17-10
Ward      4-7
Clarkson 13-6/7
Rusie     2-0
Nichols   0-1
Caruthers 0-2
total    62-37/38 *
* because I don't know how to credit a game forfeited to Clarkson when the game was tied.

Records by opponent position

finish  Clarkson Keefe Radbourn Welch
1st      26-20   30-38   27-35  22-30
         .565    .441    .435   .423
2nd      32-32   29-44   25-29  27-29
         .500    .397    .463   .409
3rd      24-23   33-21   36-27  48-23
         .511    .611    .571   .676
4th      46-20   34-24   32-31  34-25
         .697    .586    .508   .576
5th      42-17   39-24   34-18  32-30
         .712    .619    .654   .516
6th      41-19   54-23   46-20  39-20
         .683    .701    .697   .661
7th      41-22   52-20   52-17  51-21
         .651    .722    .754   .708
8th      54-11   50-25   51-14  59-14
         .831    .667    .785   .808
9-12     24-10   27-9
         .706    .750

Records against over/under .500 teams

Record Clarkson Keefe   Radbourn  Welch
.500+   118-99  147-150  141-129  129-130
         .543    .495     .522     .498
.500-   201-69  201-81   162-62   179-73
         .744    .712     .723     .710
% of career dec'n vs. .500+teams
         .446    .513     .547     .501

The following numbers are from Chris J.'s site:
run support index
Run Support Index:
John Clarkson 109.46
Tim Keefe 107.16
Ol' Hoss Radbourn 106.83
Mickey Welch 102.79

Defensive Support Index:
John Clarkson +29.9
Radbourn +15.7
Tim Keefe +15.1
Mickey Welch +5.4
Clarkson, Radbourn, and Keefe are in the top 12 all-time, while Welch is #84.

In terms of over/under achieving based on the support they have been given:
Mickey Welch overachieved by 13 wins
John Clarkson underachieved by 2
Ol'Hoss Radbourn underachieved by 5
Tim Keefe underachieved by 18.

Welch does not deserve to be excluded from the HoM. He faced more HoM opponents than anyone (Galvin may have a few more, I don't know for sure.) and performed at a much higher level than any of them head-to-head. His era+ does not compare to the others, but hopefully I will be able to start and finish my study and determine whether or not Welch pitched to the score sometime in August.
He had some of the poorer run support of the pitchers in his era as well as less defensive support, yet he still won as many games as they did. In fact, he did more with what he was given than other pitchers.
Mickey Welch deserves inclusion in the Hall of Merit.
Any mistakes are due to the fact it is almost 3:00am.

Clark Griffith
HoMers:

Clarkson  0-1
Young     1-6
Nichols   6-7
Rusie     1-3
McGinnity 2-0
Plank     2-3
totals   12-20

Records by opponent position

finish  Griffith
1st      14-20
         .412
2nd      17-20
         .459
3rd      27-20
         .574
4th      18-10
         .643
5th      18-16
         .529
6th      25-9
         .735
7th      27-14
         .658
8th      24-9
         .727
9-12     53-18
         .746

Records against over/under .500 teams

Record Griffith
.500+   112-93
         .546
.500-   111-43
         .721
% of career dec'n vs. .500+teams
         .571

Remember that during the 12 team NL 1894-99, there were some awful teams playing so it was a little easier to play .500 ball. In those 6 years, there were 42 .500 or better teams, or 7 a year. Also, in 3 of the first 4 years of the AL, there were 5 teams at .500 or better. This is a factor in Griffith's high percentage of games pitched against teams over .500. Some of you may want to adjust his totals because of this, some of you may not.
Run Support Index:
Griffith: 105.5
This is approximately the 20th best run support index for pitchers up through around 1905.
Defensive WinShares Support:
Griffith: +6.3
This is 75th all-time and 20-22nd for pitchers up through roughly 1905.
Over/Underachiever:
Overachieved by 7 wins - tied for 34th best ever.

hopefully I will be able to start and finish my study and determine whether or not Welch pitched to the score sometime in August.

No sarcasm intended, but how it is possible from looking at final scores or line scores to read the pitchers' minds and determine which ones were pitching to the score and which ones weren't?

The standard deviation due to luck for Welch's number of decisions and winning percentage is 16. If he won 13 games more than expected, that is not close to being statistically significant (statistically significant is conventionally defined as being more than two standard deviations from what is predicted).

The notion that Welch had some special ability to win games simply has no basis to begin with.

The standard deviation due to luck for Welch's W/L in #343 should actually be 11. (I forgot to carry the 2.)

What would everybody consider "easing up," "pitching to the score," etc.? John Murphy and I discussed something along the lines of if a pitcher was up by a large number of runs (5-6) in the 7th inning or later and gave up a couple of garbage runs. Numerous pitchers in the pre-Lively ball era said they would "coast" and turn it up a notch if they had to. I would analogize it to giving up garbage touchdowns or baskets in the waning minutes of a blowout. The game is over, 99% of the time, the score doesn't effect the decision, only the stats.
Theoretically, if you knew you were pitching every other game, you were up by 6 runs with 2 or 3 innings left, would you relax? - if you are Welch do you throw your curve less often? - if you are Mathewson do you throw your fadeaway/outshoot/screwball less often? There may not any significant evidence in the linescores, but if there is a pattern of teams or specific pitchers with large leads giving some of it back I think it would be important to know. Remember ERA wasn't an official stat until 1913(?).
I know we can't read players minds.
That is why I am asking the readers here this question: What would constitute to you evidence of pitching to the score?

Lots of late but unsuccessful rallies by opponents (more than in an equivalent sample of 2004 games.)

Blowouts where the starting pitcher for the losing team pitches a complete game

Those would seem to be the two situations that one could spot.

C+ defender by Win Shares, . . .
Regularly in the 2B defensive Win Shares leaders,

Strange. Does this indicate a Bill James blunder?

I don't think so. The C+ rating is based on his WS rate per defensive game played. The leader appearances are based on total DWS accumulated for the year (function of playing time in part) and list the top-5 totals in the league. Making the top-5 out of 8 (often 7 or 6 due to injury or replacement level turnover) regular 2b-men is not a significant accomplishment; more significant would be the number of years not on the leaderboard.

Doyle has 13 seasons as regular; 3 in top 3 defensively, 6 in middle two, 4 in bottom 3 (of 8) defensively. Looks like a C.

Wheat vs Hooper.

WARP-2 sees them as essentially equivalent. Long careers, little peak. Wheat has more bat (101 extra BRAR), Hooper has more glove (98 extra FRAR), Wheat has 101 more games played. For Win Shares, Wheat has the advantage of playing in the lesser league.

I can't see inducting one without the other.

Remember that during the 12 team NL 1894-99, there were some awful teams playing so it was a little easier to play .500 ball.

This is a good point & in my mind, may be a reason to downgrade him a slot on my ballot (part of the reason I had him over Welch is his high Median Opponent Winning Percentage).

The usefulness of Median Opponent Winning Percentage or Average Opponent Winning Percentage would be improved by having an expected version of the same to compare it to. Players on very good teams will expect to have low numbers and players on very bad teams will expect to have high numbers because they never play against their own team.

BP said that they already correct for this when figuring their WARP numbers. Baseball-reference "park" factors also take this into consideration. Win Shares does not. However, BR assumes no managerial bias when calculating OPS+ and ERA+; non-random opponent selection (avoiding good teams/pitchers to pitch against bad teams/pitchers or vice-versa) would require an additional correction. My guess is that BP's correction is of the same kind (that is, at the team schedule level) and does not take individual playing patterns into account. So those stats (WARP, ERA+, OPS+) have already corrected for, say Caruthers having an expected .481 AOWP.

What needs to be taken into account is how the actual AOWP deviates from the expected AOWP.

W-L record by Run Margin in game started:

        Welch         Keefe         Galvin
    0:  20 ties       13 ties       14 ties
    1:  76-43 .639    64-68 .485    78-70 .527
    2:  47-41 .534    48-34 .585    52-55 .486
    3:  49-26 .653    50-36 .581    50-45 .526
    4:  36-28 .563    43-31 .581    40-33 .548
    5:  37-23 .617    27-17 .614    30-36 .455
 6-10:  54-39 .581    93-42 .689    89-56 .614
11-15:  15-10 .600    20-3  .870    19-13 .594
16+  :   3-2  .600     5-0 1.000     5-2  .714

Total: 317-212 .599  350-231 .602  363-310 .539
Close: 172-110 .610  162-138 .540  180-170 .514
Close:  55.0%         52.7%         53.0%
Tight:  25.3%         24.4%         23.6%

        Radbourn      Clarkson      Caruthers
    0:  8 ties        13 ties       4 ties
    1:  58-47 .552    60-45 .571    34-23 .596
    2:  55-34 .618    38-30 .559    36-22 .621
    3:  37-32 .536    36-30 .545    21-12 .636
    4:  31-19 .620    46-19 .708    28-13 .683
    5:  35-18 .660    35-12 .745    17-9  .654
 6-10:  68-34 .667    94-35 .729    57-17 .770
11-15:  14-3  .824    19-4  .826     7-3  .700
16+  :   6-3  .667     2-2  .500     8-4  .667

Total: 304-190 .615  330-175 .653  208-99 .678
Close: 150-113 .570  134-105 .561   91-57 .615
Close:  54.0%         48.6%         48.9%
Tight:  22.5%         22.8%         19.6%

Close is when margin is 1-3 runs. Close % is percent of starts that were close. Tight % is percent of starts that were ties or 1-run games.

Make of it what you will.

Great work jimd.
I am not surprised Caruthers and Clarkson have the lowest percentage of "close" games because their teams were so dominant at times. Caruthers and his Browns and Clarkson with Boston and Chicago. Also, Caruthers and Clarkson have the highest RunSupport Index so it makes sense they have fewer close games.
Also, while Welch's difference/ advantage in the percentage of close and tight games is small it is not unexpected as he has the second lowest RSI of the 6 and the second lowest defensive support of the 6. Of course, Galvin is the lowest in both measures but has a smaller percentage so I'm probably just talking out of my hat.
More on Welch. From the numbers, it appears as though Welch was the pitcher to have in a close game. He was the only one to improve his winning percentage in close games while the other pitchers declined by .062, .025, .045, .092, and .063. So was Welch pitching to the score or did he just win with what was given him? If you gave him 4, would he give up 3 whereas the others would give up 2? I guess I will have to check and see when some of these pitchers gave up their runs.
Great work again, jimd

Thanks for the kind words, Kelly.

Errata above: In 16+ games, Clarkson was 2-0, Caruthers was 8-0.

W-L record by Game Margin in game started:

        Composite:
    0:  72 ties
    1:  370-296 .556
    2:  276-216 .561
    3:  243-181 .573
    4:  224-143 .610
    5:  181-115 .611
 6-10:  455-223 .671
11-15:   94-36  .723
16+  :   29-7   .806

Total: 1872-1217 .606
Close:  889-693 .562
Close:   52.3%
Tight:   23.3%