Applying it historically would require a study of the evolution of bullpen usage over time

... which, coincidentally, I've been doing for the last 18 months or so.

-- MWE

Mike, roughly when did the 'fireman' emerge as a bullpen feature? And when did the Fireman Era end? Who was The Last Fireman?

Isn't the distinguishing characteristic of the modern era that relievers are brought in at the start of innings?

Isn't the distinguishing characteristic of the modern era that relievers are brought in at the start of innings?

I'm sure that's one of them, but the bigger would be that roles are defined by the inning in which you pitch, rather than by how much "trouble" there is.

I made a stab at trying to answer my 'Last Fireman' question myself. Mike has posted some data about blown leads in Primate Studies. Sorting it by number of times a pitcher was brought in with a lead, and by those leads lost after the first inning of work puts Lee Smith as the latest man chronologically near the top of both.

Mike, roughly when did the 'fireman' emerge as a bullpen feature?

By 1960 the fireman was well-established. Before about 1955, with the exception of the teams that Branch Rickey directly affected (Dodgers, Pirates, Cardinals) and Casey Stengel, there weren't that many teams who used them. Stengel's success in the AL, and Brooklyn's in the NL, drove the fireman model from what I can tell.

And when did the Fireman Era end? Who was The Last Fireman?

The Fireman Era ended over the 15-year period between 1975 and 1990. That was an era of rapid role changes in reliever usage. Lee Smith, mentioned in #104, was probably the last career reliever who could have been said to be used as a fireman.

Isn't the distinguishing characteristic of the modern era that relievers are brought in at the start of innings?

Yes. That and reducing the number of innings that you expect any reliever to pitch in any one game, so that you can have any reliever available at any time. It has become rare for a reliever to pitch more than two innings in a game, absent extra innings.

I'm sure that's one of them, but the bigger would be that roles are defined by the inning in which you pitch, rather than by how much "trouble" there is.

Well, that how we perceive it. But the determination that it was better to have a fresh arm to start an inning came first - then the assignment of those arms to specific innings came later.

-- MWE

I'm not at all sure that the fireman concept has dropped out of baseball. Isn't a LOOGY essentially a fireman, someone you bring in when there already is trouble? I realize that "trouble" in this context means a hot lefty hitter rather than necessarily men on base, but the concept is the same.

I am old enough to have listened to radio during the time period when the idea of bringing a reliever in to start an inning really took hold. The reasoning given by the managers amounted to 2 things:

1) If I (the manager) think there is a chance that my current pitcher is about done, why not just bring in the next guy to start the inning instead of running the risk of the current guy getting into trouble, which increases the chance that I'll end up losing the game?

2) Pitchers HATE it when they let men on base and then some other pitcher allows the men to score. They think "Hell, I could have let those guys score."

It's sort of like what happened to Tony La Russa's experiment with using three pitchers in each game, replacing them at the start of the 4th and 7th inning. I actually got to ask him about that during my tenure as a newspaper writer, and he said that the trouble was that no one wanted to start, because the rules of baseball require you to pitch 5 innings to get a win, but you can lose the game without getting anyone out. No one wanted to negotiate their next contract with a W/L of 0 and 11. - Brock

The Fireman Era ended over the 15-year period between 1975 and 1990.

Fascinating. I'm surprised at the earlier date, but remembering Bruce Sutter in 1977 makes me suspect I shouldn't be. I always conceive of it as a relatively sudden process, taking place during 1985-90.

Thanks, Mike!

When I think of a "fireman" reliever I think of someone like Mike Marshall or Gossage - best reliever brought in for 1+ innings. A LOOGY is brought in strictly for platoon advantage and dumped shortly thereafter. Fireman evolved into "setup reliever" but he's now the 2nd or 3rd best reliever. The best guy got moved to the closer role.

DL - I understand your definition, and it makes just as much sense as mine. Using your definition, I have a memory (please correct me if it's off) that the end of the fireman role - the bringing in of your best reliever for more than the ninth inning - took a huge jump when Dennis Eckersley, who was with Oakland and tearing the league apart as a closer, said that it was important to keep his yearly innings to down around 60, certainly under 70. This produced a pretty sudden and generally accepted change of the role to one inning only. I remember this pretty strongly, and remember it as being all over the sports media, with managers changing their usage very quickly (as manager movements go) so I hope I haven't magnified it in my mind somehow. Eck, if I remember right, was also very instrumental in changing the perception of the effects of Coors Field from "the thin air doesn't provide as much resistance to fly balls" to "the thin air doesn't provide enough resistance to let a curve ball curve properly." As a consequence, I remember Eck as being the very most influential pitcher of the last 40 years or so, in terms of changing general perceptions of what actually happens in the game. If the constitution allowed credit for this sort of thing, which I don't think it does, Eck would end up moving way up in my various HoM lists.

Hey! I just heard over the TV that the Veterans' Committee has just elected Whitey Herzog into the Hall of Fame. I guess they don't get them all wrong.... And he's going in with Doug Harvey, who is a pretty good pick himself, if you're inducting umpires.- Brock

I have a memory (please correct me if it's off) that the end of the fireman role - the bringing in of your best reliever for more than the ninth inning - took a huge jump when Dennis Eckersley, who was with Oakland and tearing the league apart as a closer, said that it was important to keep his yearly innings to down around 60, certainly under 70.

As I have pointed out before, Eck wasn't the real driver behind the one-inning closer. He was generally used a little less often before the ninth than was the typical closer, but he was still used a lot before the ninth. From 1988, when he became a closer, until 1993, Eckersley had 30-35% of his relief appearances start prior to the ninth inning (the norm for that period was around 37-40%). The number started trending downward in 1992, and by 1994 Eck was down to 20%.

The first real one-inning closers were Tom Henke and Gregg Olson, both in 1991 and both in part because they were recovering from injury. When Phil Garner took over the Brewers the following year, he put Doug Henry in as his closer and pretty much limited him to one inning stretches; as you may remember, the '92 Brewers were surprisingly successful. Butch Hobson, who took over Boston that year, did the same thing with Jeff Reardon with markedly less success, and Joe Torre (!) did the same thing with Lee Smith in St. Louis. By 1993, you saw a few more managers pick it up - notably Philadelphia's Jim Fregosi with Mitch Williams. By 1994, virtually everyone, including TLR, was doing it that way, and Lee Smith under Johnny Oates (one of the pioneers with Olson) in Baltimore became the first closer to enter every appearance in the ninth inning or later.

-- MWE

Fascinating. I'm surprised at the earlier date, but remembering Bruce Sutter in 1977 makes me suspect I shouldn't be.

I think the real eye-opener was what happened to Mike Marshall after his 106-appearance season in 1974. Marshall went on the DL in May of the following season after being hammered for nine runs in two innings by the Pirates, and while he pitched reasonably well when he came back he wound up 9-14 overall with eight blown saves in 21 opportunities. That convinced some people that ace relievers couldn't continually be asked to handle every tight late-inning situation. Sutter added some additional fuel to that fire, but the trend toward fewer innings, especially when trailing or tied, for the ace reliever had already begun by that point.

One thing that I'm finding is how little the conventional wisdom about who did what when is actually borne out by looking at real usage patterns. We tend to give TLR a lot more credit for shaping modern usage of the bullpen than he merits, when the only thing that he can really be said to have pioneered to any extent is the LOOGY (Honeycutt).

-- MWE

You'd have to be a 100% pure peak voter to dismiss Palmeiro.

I agree. I've been tinkering with the formulas and I'm getting closer to consensus. Still more pitchers than the HoM (by design) and I still like Bob Johnson and Ben Taylor. Palmiero moves up to 12th and the SS move down but Rizzuto is still on-ballot. Palmiero, Jake Beckley and Ben Taylor have a lot in common.

Big movers up: Tommy John, Don Newcombe, Bucky Walters (all still off ballot)
Big movers down: Tommy Leach, Dick Redding (he's really a peak candidate - Dwight Gooden with a less prononounced decline)

"I think the real eye-opener was what happened to Mike Marshall after his 106-appearance season in 1974."

And of course the more amazing figure was the 208.3 IP and 15 W - ALL in relief.

That must seem impossible to you youngsters under 40.
:)

In 1973-74, his combined numbers were 29-23 with 52 SV, 198 G (all in relief) and 387.1 IP, with a 2.53 ERA.

The 1974 IP total was 29th in the league, behind 28 SPs, and 3rd on the Dodgers.

Interesting that this was the same franchise that had just used up Koufax 7-8 years earlier (of course, both reached the postseason).

Part of the work that I have done was to identify the key reliever - the guy who came into the key late-inning situations with a lead with the expectation that he would finish the game - on each team for the years covered by Retrosheet (1952-2009). Some teams before 1960 had none; some teams since 1960 have had more than one who split the workload; but since 1960 nearly every team has had one.

For the teams that had one, and only one, identifiable key reliever, I did a further breakdown by the inning in which that pitcher came into the game.

The rate of entry for top relievers was fairly consistent throughout the sixties. 70% of the time, give or take, they entered games prior to the ninth inning. The rate dropped in the early seventies, as starter innings actually went back up, expansion thinned out the quality of the bullpens, and teams experimented with new bullpen models, but by 1974 the ace relievers were back up over 70%. After the 1977 expansion, ace relievers started coming into games less often once again, and the trend accelerated all through the 80s. By 1988, ace relievers were coming into the game before the ninth less than half the time. By 1993, that figure was down to about 30%. Within 5 years, that figure was down near 20%. In 2007 and 2008, the figure was near 10%, but it jumped back up in 2009 for reasons I haven't investigated yet - it had been hovering around 15% for several years prior to 2007, and it may be that 2007-2008 were aberrations, or it may be that we're starting to see a move back toward increasing the closer's responsibility.

-- MWE

I've done some more reworking of my pitching system over the past couple of weeks (mostly, incorporating Totalzone for defensive support), and here are some things I've noted over this.

- According to this, Tiant was really helped by his defenses, which lowers my ranking of him (with Reuschel moving in the opposite direction). He may probably still make my ballot, but definitely not in an elect-me position.

- As for 19th century pitchers, Jim McCormick may be the most interesting one left behind by now.

- Not that it matters for HOM elections by now, but Red Faber and Eppa Rixey don't look too well with this revision.

- An opposite of long career pitchers from the '20s? High-peak pitchers from the '80s. Gooden and Hershiser become ballot candidates now.

- Cone is now my top backlog pitcher. I've already run Brown, and he's comfortably above the backlog, so no "weirdness" there.

Passes the smell test, Juan V.

Good article here about Art Pennington (living NgL player) playing his Strat card agains Satchel Paige's strat card. Some good tidbits:

"Everyone knew Ted "Double Duty" Radcliffe could throw hard, Pennington says. What they didn't know was that he was cutting the ball."

"Simkus found that Gibson hit 10 home runs in fewer than 40 games at Washington's spacious Griffith Stadium. The entire American League didn't hit that many there in 76 games against the Senators that season." - there's more on Simkus data in the article

[each time the big league club called, it wanted another player, not Pennington, who had married a light-complexioned Spanish woman a decade earlier while playing in Mexico.

"The wife. That was the whole thing," Pennington says. "They actually said, 'Would you leave your wife?' I said, 'Man, I wouldn't leave my wife for all of baseball.' "]

Some interesting tidbits on Kevin Appier vs Dave Stieb. "Expected Leads Lost" is a measure of the closeness of the games; it is based on the size of the lead and the run-scoring environment, and is derived from Dave Studenmund's WPA spreadsheet. A pitcher is given credit in my system for an actual lead lost if, when he left the game, the probability of the tying run scoring exceeded 50%.

Appier:

Innings Started: 2662
Innings Started with a Lead: 1117 (42.0%)
Expected Leads Lost: 149 (13.3% of all leads)
Actual Leads Lost: 113
Pct Actual/Expected: 75.8%

Stieb:

Innings Started: 2931
Innings Started with a Lead: 1207 (41.2%)
Expected Leads Lost: 158 (13.1% of all leads)
Actual Leads Lost: 140
Pct Actual/Expected: 88.6%

Both Appier and Stieb were well below the median (46%) in terms of the percentage of innings that they started with a lead in the group of 38 HOF starters and HOF-potential starters that I have been working with and for which we have good portions of their career in Retrosheet event files, and they were both above the median (12.6%) in terms of the expected percentage of leads lost (e.g. they were both pitching with smaller leads and/or in higher run evnironments, which made it somewhat more likely that the lead would be lost). Stieb's 88.8% of actual leads lost vs expected leads lost was one of the worst of the group of 38; Appier was above the median (76.5%).

Pedro Martinez, by the way, is by far the best pitcher in this group at working with a lead. He has started 1317 innings with a lead in his career, would have been expected to lose 183 of those leads (13.9%) but has actually been responsible for losing only 105 leads (57.4%). Roger Clemens, Bret Saberhagen, and Randy Johnson have also done well, all under 70% actual vs expected. Guys who have done poorly (over 90%) are Phil Niekro, Jack Morris, Jim Kaat, and Catfish Hunter.

-- MWE

Not really surprising that Pedro, Clemens, Saberhagen and Randy Johnson would be better at holding leads than Morris, Kaat and Hunter. They're better at run prevention than Morris, Kaat & Hunter also.

You avoided listing Blyleven - any particular reason for that?

What do these lists look like when you normalize for the pitcher's RA?

I insist that I think this is quite a kooky way to evaluate pitchers, as if pitching in tie games or when trailing simply doesn't matter. If you derive these numbers from WPA, Mike Emeigh, then why not just use WPA over replacement itself?? That would give you the full measure of true in-context value that you're looking for, not this sliced-and-diced subset that ignores what, 60% of innings, some of which (ties or trailing by 1) have quite high leverage?

That would give you the full measure of true in-context value that you're looking for, not this sliced-and-diced subset that ignores what, 60% of innings, some of which (ties or trailing by 1) have quite high leverage?

I don't care "when" a pitcher loses a lead, only "that" he loses a lead - I don't use WPA for this purpose because in WPA "when" it happens is more important than the fact that it did happen. The reason I do it that way is simple: protecting a lead, at any point in the game, is the most important thing that a pitcher can do to help his team. If he's behind, he helps his team by keeping it close, but not nearly as much as he does by keeping it ahead.

In almost all cases, pitching with a one-run lead is a higher leverage situation (usually by quite a bit) than pitching when tied or trailing by a run, controlling for run environment, inning and base-out situation. In the vast majority of cases, pitching with a two-run lead is also higher leverage than pitching while trailing by a run or tied, and pitching with a three-run lead is almost always higher leverage than pitching while trailing by a run, although not while tied. Tie games "seem" like they should be higher leverage than pitching with a one-run or two-run lead, but that's not normally true until the late innings (and even then tie games are virtually always lower leverage than pitching with a one-run lead).

That doesn't mean that pitching when tied is trailing is *un*important, only that it is *less* important than pitching with a lead. If these guys didn't do well at holding leads and couldn't keep games close when they were trailing, they wouldn't have gotten into the HOF consideration set in the first place. Once they are in the HOF consideration set, then I look at "when" they prevented their runs.

To answer the other question: Blyleven ranks just above Stieb on the percentage scale, about even with Gaylord Perry and Jim Bunning, neither of whom would be in my personal HOF either.

-- MWE

To answer the other question: Blyleven ranks just above Stieb on the percentage scale, about even with Gaylord Perry and Jim Bunning, neither of whom would be in my personal HOF either.

Out of curiosity, how big is your personal Hall of Fame? Is it the same size as the actual Hall of Fame? I ask, only because I have a hard time understanding (given my own preferences) how 5350 IP of 117 ERA+ ball could fall outside of the Hall of Fame, as currently sized.

Your personal hall of fame is boring, small-hall guy. I don't like your silly criteria either. A game starts tied, if a guy pitches 7 shutout innings while his team fails to score that equals no value in your system. Holding a tie should be at least as valuable as holding a lead.

You can get rid of the "when" in WPA by dividing by the leverage.

Interesting that this was the same franchise that had just used up Koufax 7-8 years earlier (of course, both reached the postseason).

What is fascinating to me is how Mike Marshall argues that he could handle certain workloads because he knew/knows stuff about pitching mechanics that no one else does (and he won't tell you unless you pay him$)- but in reality Marshall COULD NOT handle those workloads (not for long any way).

WRT Koufax & Marshall- Drysdale too, 300+ IP 4 years in a row. led league twice, began breaking down, career over at 32

Osteen had a 321 IP season for them
Messersmith threw 321, broke down and was done at 32/33
Singer threw 315, broke down, was fixed up after a few years, threw 315 again, broke down again and was never the same again

Sutton- never threw 300+ IP, 293 was his top, but he could throw 250ip a season for frakking ever...

The Alston/Lasorda Dodgers? There is something to that, a bunch of guys threw 300+ IP for them, and while some could obviously handle more than a year or two (Drysdale)- I think he lost years of his career- Singer's career looks like it was wholly gutted by overuse

I still just don't get it at *all*. Since you yourself recognize that innings in which a pitcher's team is not ahead also count, then why not just use either WPA or WPA/LI to weight them accurately, instead of giving them a weight of zero by only considering situations when a P had a lead?

For those who missed it, Joe Posnanski discussed The Hall of Merit and Fame in a December 28 blog post.

Just curious. What kind of league adjustments do people have for the duration of Brown's career? His career can be fairly divided up into an AL half (good but not great)and a NL half (short but great).

With no league strength adjustments, I have his RA+ equivalent record as 120-58 in the NL and 96-89 in the AL.

If it's elect three, I'd go unquestionably with Bagwell and Palmeiro and put Brown as a close third choice over Walker. I know the argument that "a great hitter takes advantage of his home field," but Walker's numbers are so inflated by Coors that IMO it pushes him just below the margin, at least compared to the other three. And since this is the Hall of Merit and not the Hall of Fame, I don't pay any attention to Palmeiro's steroids.

Is the rep higher for AL pitchers in Dan R's WAR to account for the fact that they don't have to bat? It looks like I'm giving too much credit to AL pitchers for their zero hitting contribution.

I've been interested in sabermetrics and the Hall of Merit for years, but this is my first time posting here. I would like to vote in next year's HoM election, and I've come up with a preliminary list based on a rough mixture of career and peak rWAR (I used career total, top seasonal total, top 3 seasonal totals, and top 5 consecutive seasons):
1. Bagwell
2. Brown
3. Walker
4. Reuschel
5. Mullane
6. Bando
7. Tiant
8. Buddy Bell
9. Olerud
10. Appier
11. Cedeno
12. B. Bonds
13. McCormick
14. W. Davis
15. Silver King

How do you guys rate Negro Leagues players? Do you guys have a link to a reputable site with Negro Leagues statistics? Right now I've left Negro League players off entirely due to a lack of information.
Also I've read in some of the comments about Gavvy Cravath's outstanding minor league record. Do you guys have some system for MLEs in the early 1900s? As far as I can tell, he was only outstanding for maybe 4 seasons (based on Baseball-reference's admittedly incomplete minor league stats).

The main comment: the back threads of the HoM are a very rich resource. Go to the front page of the HoM and click on "Important Links."

As for rating Negro League players: we've depended a lot on the statistical legwork of some of our members (such as KJOK) and friends (such as Gary A), and we've developed our own MLE systems - notably Chris Cobb and Eric Chalek. The results of that are scattered in a lot of places - the thread entitled "All Time Negro League All Stars," the threads of specific Negro League players (which you can find catalogued under "Links to discussions of Negro League players"), the "yearly" discussion thread for the years in which such candidates were commonplace.

The issue of Cravath in the minor leagues got pretty thoroughly aired on his own thread.

Now, as for that prelim:

You seem to have a special liking for three categories of players: 60's-70's-80's 3B, 60's-70's-80's CF, and 1880's pitchers. Been a long time since we've had a Silver King sighting around here. (Not your great-great-grandfather, is he?) By vote totals, the highest-ranking remaining 1880's pitcher is Mickey Welch. I always had McCormick as the best of the lot but haven't voted for him in a long, long time. Do be aware that the AA and the NL weren't always equal-strength leagues. That affects the ranking of both Mullane and King.

Is the rep higher for AL pitchers in Dan R's WAR to account for the fact that they don't have to bat? It looks like I'm giving too much credit to AL pitchers for their zero hitting contribution.

DL, I give pitcher hitting credit based on above/below average what other pitchers hit. This makes sense because replacement level hitting for pitchers is what an average pitcher hits (since pitchers aren't chosen for their hitting).

So the average NL or pre-1973 AL pitcher gets a zero, and likewise for DH era AL pitchers.

Alex - have you read through the entire HoM thread on Chone's WAR? I think you should.

Have you figured into your data pitcher hitting and fielding? Who is 16-30 on your list?

I'd take the presence of Mullane high on Alex's list as at least a hint that pitcher hitting might have some influence.

DL from MN--in my system, pitchers are fully penalized for their hitting below league average, but then get that value back in the replacement column, where they are compared to an average-hitting pitcher in the same number of PA. So a typical good-hitting pitcher might be 0.0 PWAA, -0.5 BWAA, and -3.5 Replc (which breaks down to a replacement pitcher being 2.3 pitching wins below average and 1.2 batting wins below average), for 3.0 WARP total. AL pitchers are simply a special case/subset of this: they have 0 PA, so both they and the replacement pitchers they are compared to have 0 BWAA.

Updated prelim ballot - I have attempted to correct for post-1973 AL pitchers getting extra credit. This many years in and I still make mistakes like this, partly because I'm not diving deeply enough into the numbers. It doesn't change a whole lot. I've also been fiddling with the positional average information rippling through my PHoM. Finding lots of interesting things.

1) Bagwell - no change here
2) Brown - no change here either. Still comparable to Jim Palmer and Fergie Jenkins.
3) Larry Walker
4) Tommy Bridges - he deserves war credit, tremendous in the postseason also
5) Rick Reuschel
6) David Cone - the AL hitting moved him down a tick
7) Urban Shocker
8) Phil Rizzuto - climbs a bunch with the WAPA numbers
9) Gavy Cravath
10) Rafael Palmeiro - a tick above Will Clark and Keith Hernandez due to a longer career
11) Luis Tiant - another AL pitcher hitting casualty
12) Bus Clarkson
13) Bob Johnson
14) Ben Taylor - if you like Palmeiro and Jake Beckley give Ben Taylor a second look
15) Bert Campaneris - probably won't last though, he has some unusual value
16-20) Bucky Walters (big move up), Dave Bancroft, Lave Cross, Norm Cash, Johnny Pesky
21-25) Don Newcombe (++), Dick Redding, Dave Concepcion, Babe Adams, Tommy Leach
26-30) Dizzy Dean, Jack Fournier, Wally Schang, Wilbur Cooper, Tommy John
31-35) Dizzy Trout, Kevin Appier (makes more sense now), Hilton Smith, Leroy Matlock, Tony Mullane
36-40) Jim McCormick, Dwight Gooden, Dolf Luque, Vern Stephens, Fred Dunlap

One really interesting guy - Pie Traynor. He gets a 48.7 WAR with 9 fielding wins above average. This does not match his reputation. Robin Ventura was +15, Brooks Robinson +22, Graig Nettles +16, Ken Boyer +12, Eddie Collins +13, Jimmy Collins +19, Ron Santo +11. The anectdotal evidence has Traynor more valuable than just about anyone in his era defensively. This would suggest that the WAR calculators are systematically underrating him. If I add 5 WAR to make him a gold glove fielder, like his reputation suggests, he vaults from 54th to 20th on my list. Still not quite enough to vote for him but much closer to his reputation.

I should add that the adjusted Pie Traynor ranks higher than Ken Boyer. Swapping Ken Boyer for Pie Traynor in our all-time rankings would better match our inductees to conventional wisdom.

Swapping Ken Boyer for Pie Traynor in our all-time rankings would better match our inductees to conventional wisdom.

I agree. I still think Boyer is the worst HoMer position player (with Lemon the worst HoMer pitcher). I know someone has to be that guy, but I don't think he's a great choice for the 'honor'.

If we don't elect Boyer, I think Dick Redding would be in. We would have elected Fox in 1991 instead. So in 1997 we would have elected Beckley instead of Fox and then in 1998 Redding would have gotten in instead of Beckley.

If we also don't elect Lemon, I think Bucky Walters would be in. We would have elected Rixey in 1967, which means Mackey would have gone in in 1968, instead of 1975. Willard Brown would have went in in 1975 instead of 1976, Sisler in 1976 instead of 1979, Mendez in 1979 instead of 1985, Pierce in 1985 instead of 1987, Childs in 1987 instead of 1988, Wynn in 1988 (ahead of Boyer) instead of 1996, (Fox as mentioned early, over Boyer in 1991), Roush in 1996, and Bucky Walters in 1997.

I'd much rather have Redding/Walters than Boyer/Lemon. Obviously I'm in the minority.

Alex King:

I suppose DL from MN and I came across like we were assigning you homework. We do want to welcome your efforts. Of course, it is a long time from now until that next vote - plenty of time for you to get your questions answered, and to pose more questions if necessary.

I'm with you on Lemon as the worst pitcher elected, though I'm also not a fan of Rollie Fingers.

I still like Boyer more than Nellie Fox. I'm not a big fan of Pete Browning or George Sisler either. I really dislike Willard Brown, I see him as Andre Dawson minus the defense which to me is a really weak candidate. Boyer's not great but he's not the bottom of my barrel.

OCF/140:
It's fine; I appreciate all the links and advice you guys gave me.

DL/134:
Yes, I have accounted for pitcher hitting, using Chone's WAR. However, if the pitcher contributed less than +/- 2 hitting wins over his career, I ignored pitcher hitting (since this is only a rough first ranking). Later I think I'll go back and add pitcher hitting for the rest of the players, although it should only make a difference at the margins.
Also, my 16-30:

16. Cone
17. Ventura
18. Gooden
19. Cicotte
20. Tanana
21. Cash
22. Willis
23. Finley
24. Charlie Buffinton
25. Hershiser
26. Rizzuto
27. F. Chance
28. Shocker
29. Wilbur Wood
30. Dizzy Trout

In addition, I'm going to re-investigate Tommy Bridges (war credit) and Gavvy Cravath--right now, I have Bridges at 49 and Cravath at 61.

As for the 1880s pitchers:
1. I disregarded any stats in the Union Association of 1884. This affects only McCormick as he posted 8.8 WAR in the 1884 UA.
2. I gave a slight boost to the NL pitchers. I'm going to calculate the replacement levels Chone used in his WAR to see if they are reasonable--he may not have incorporated league adjustments for this time period. Also I might throw in the UA stats, adjusted with a much lower replacement level.
3. I multiplied all the peak-value stats I considered (top season, top 5 consecutive, and top 3 seasons) by 2/3. I don't really have any reason for 2/3 other than it seemed to feel about right: the adjusted peak values for the 1880's pitchers were similar to the real peak values of other pitchers.

I don't know how you can get all the way to #30 without hitting Palmiero. I could tell immediately that you like Chone's WAR because you have the 70s 3B up high and Phil Rizzuto down low.

Whoops. I made 2 lists, one for hitters and one for pitchers, and when I put them together I completely missed Palmeiro. I would put Palmeiro 6th right now but he could pass Mullane when I re-assess the 1880s pitchers.

I agree Boyer is pretty bad, but that doesn't mean it would have to be a piece of Pie in his place.

Thanks for the further explanations of your methods, Alex King. I would, however, strongly encourage you to take a look at the eligible Negro League players. It's not enough to say "Right now I've left Negro League players off entirely due to a lack of information." We are required to be fair to all eras, and that includes players who were prohibited from playing in the official major leagues based on the color of their skin. Assess the information we do have.

Based on the players you do have on your ballot, you may just find a number of Negro League players to your liking. Cannonball Dick Redding has a similar profile to many of the 1880s pitchers, with a lot of big innings and big years. Or you might surprise us and become the lone voice for someone like John Donaldson, a pitcher who could hit so well that he became an outfielder for the latter half of his career.

The additional caveat on Chone's WAR is you need to adjust it for schedule length. We're pretty adamant that a "pennant is a pennant" around here.

As far as Negro Leaguers go, my top NGL guys are Ben Taylor (see Beckley, Palmeiro) who is in the HoF, Dick Redding - who missed the HoF cut but was in the consideration set, Bus Clarkson - part of the lost generation, Hilton Smith (HoF), Leroy Matlock and Luke Easter. Others seem to like Bill Monroe more than I do.

I would like to see a "Chone" perspective around here, however, I think the shortcomings in that WAR for HoM voting (primarily schedule adjustments) require a little tweaking.

BTW, meant to say welcome aboard Alex!

Dan - can you clarify if an average hitting pitcher (average amongst pitchers) in a season gets a net of zero in your WAR?

I believe this is how it should work (along the same principle that replacement for fielding is an average fielder, not a bad fielder), but want to make sure that's how you are doing it.

Thanks!

Chris/146:

Definitely, I agree. I hadn't considered Negro Leagues players due to the lack of easily available evidence; I will look at the Negro League players' threads on this site and I may also do some additional research.

DL/147:

When you say "you need to adjust it for schedule length," do you mean pro-rating 154-game seasons to 162-game seasons (and also adjusting war-shortened, strike-shortened, and 1870's/1880's seasons) ?
Also, I presume that you guys don't correct for schedule length for early pitchers (otherwise we'd have them pitching 1200 innings a season).

Yes a pennant in a 154 game season is worth the same as a pennant in a 162 game season but it takes more wins (WAR) to accomplish in the 162 game season. Most here also typically adjust for strike seasons and other shortened seasons. Pre-rotation pitchers are harder to adjust, their IP/season is partly cancelled out by their shorter number of seasons. Comparing pitchers across eras is one of the trickier things to do. Adjusting by the typical workload of an average pitcher is helpful.

That's correct, Joe Dimino. An average-hitting P would be docked about 1 win in the BWAA category, but the replacement pitcher he is compared to is also below average by the same amount, so -1 -(-1) = 0.

Hey guys, I have been away from the HOM for almost two years (maybe more?) and I would like to get back in. I am not sure if I would have to resubmit a ballot or anything so I am asking here.

Also, I have some other questions about stats and where to get them. Previously I used mostly Win Shares and WARP and while I know where to get WARP, does anyone know of a place to get win shares online (especially for after the 2002 season as that is all the farther that my edition of Total Baseball goes.

Also, where can I find Dan R's WAR numbers? I am thinking of adding in either Dan R's number or CHONE's.

For those who may forget i am a bit of a peak voter (though I usually scored high on the consensus ranking). My peak system takes a certain level (it was 25 WS and 7 WARP) as a baseline and counts up what a player achieved over that baseline in each seasons. Therefore peaks can be short and high (Jennings and KEller) or long (Billy Williams or AL Kaline) and still be valuable. I like this method more than the best3/5/7 model of measuring peak. I do teh same with 'average' performance as well to measure Prime and look at career values.

Dan R's data should be available on the Yahoo HoM website. Both Chone and Dan R WAR have a higher replacement level than Win Shares or WARP.

Mine are available at the link at the top of my WARP thread.

Welcome back, jscmeagol (did you drop an H from that at some point, or am I misremembering?)! A fellow peak voter has missed you. :)

I have read through the Gavvy Cravath thread and I am calculating WAR for Cravath's AA and PCL years based on Brent's MLEs (though I tweaked them slightly). I should have my numbers for Cravath posted tomorrow night. Right now it looks like he will come in around #30 in my rankings, although the adjustments I still have to do will push him upwards. In addition, he should move up a few spots when I correct for season length.

Using Brent's MLE's (third edition) that he provided in the original Gavvy Cravath thread, I calculated Cravath's WAR by imitating Sean Smith's method.
First, I altered Brent's numbers slightly. I found that his projected walk rates for Cravath were unreasonably high. When combined with Cravath's major league numbers (1909 excluded) we obtain the following list:
1903 0.091
1904 0.096
1905 0.112
1906 0.116
1907 0.118
1908 0.116 (majors)
1909 0.130
1910 0.147
1911 0.155
1912 0.094 (majors)
1913 0.093 (majors)
1914 0.137 (majors)
1915 0.138 (majors)

It seems unreasonable to assume that Cravath had the best walk rate of his career before he became a major league star. Furthermore, Cravath's 0.093 walk rate in 1100 PA's between 1912 and 1913 shows that he wasn't a particularly patient hitter at this time; therefore I don't think there is any reason to assume that he was unusually patient before 1912. Yes, he did post an 0.116 walk rate in 1908, but that came in only 300 PA. The most likely scenario is that the 0.116 BB rate in 1908 was a fluke and Cravath's true walk rate was closer to the 0.093 he displayed in 1912 and 1913.
As a result, I simply set Cravath's walk rate at 0.100 for his entire minor league career, splitting the difference between his 1908 walk rate and his 1912-1913 walk rate. Then I adjusted at bats and hits in order to keep Cravath's total plate appearances constant.
Using these modified MLE's, I calculated Cravath's wOBA and then converted it into linear weights using the formula ((wOBA-lg_wOBA)/1.15)*PA.
Lastly I estimated the runs-to-wins conversion by using a linear regression of runs/wins converter and R/G over Cravath's career. The regression line I obtained had an r^2 of 0.96, so I'm reasonably confident that my estimated runs-to-wins converters are accurate.
Assuming his baserunning and fielding to be 0 for his entire minor league career, I obtained the following win values:
1903 0.7
1904 1.4
1905 1.3
1906 2.3
1907 3.7
1909 3.5
1910 5.0
1911 6.2
Of course, we can try to estimate Cravath's fielding and baserunning. For his career, Cravath was a -11 baserunner. However, his minor league seasons occurred before his major league career and therefore we can assume that minor=league Cravath was a better baserunner than major-league Cravath. We also know that in 1908, Cravath was worth 0 baserunning runs. Therefore, I decided that Cravath contributed 2 runs on the bases in 1903 and 1904, 1.5 in 1905, 1 in 1906, and 0.5 in 1907. He starts as a moderately good baserunner and gradually becomes average by the time he made the major leagues for the first time. For 1909-1911, I set Cravath's baserunning as average--he contributed 0 runs on the bases in 1908 and -1 in 1912, so he was probably average in between.
We also have some fielding data for Cravath's minor league career. From Brent:

For 1903, ranked 4th in fielding percentage of 6 full-time right fielders.
For 1904, fielding record shown is for right field only; record for all positions is 341 PO, 55 A, 30 E. Reach Guide does not indicate what his other position(s) were.
For 1904, ranked 2nd in fielding percentage and 1st in assists of 6 full-time right fielders.
For 1905, ranked 9th in fielding percentage and 5th in assists among 19 outfielders with 100+ games.
For 1907, ranked 1st in fielding percentage and 3rd in assists of 10 outfielders with 100+ games.
For 1909, ranked 15th in fielding percentage of 21 outfielders with 100+ games.
For 1910, ranked last in fielding percentage of 22 outfielders with 100+ games.
For 1911, ranked 15th in fielding percentage of 20 outfielders with 100+ games.

I translated these rankings into defensive runs above/below average:
1903 -5
1904 5
1905 5
1907 10
1909 0
1910 -10
1911 0
For 1909 and 1911, 15th in fielding percentage doesn't seem too good. However, remember that Cravath is being compared against 7 CF as well as 7 RF and 7 LF. Therefore, he is about average for a corner outfielder.
Cravath's updated WAR, with fielding and baserunning:
1903 0.4
1904 2.1
1905 2.0
1906 2.4
1907 5.0
1909 3.5
1910 4.0
1911 6.2
Here is Cravath's career WAR, including his actual major league numbers:
1907 5.0
1908 1.9
1909 3.5
1910 4.0
1911 6.2
1912 1.8
1913 5.2
1914 3.7
1915 7.2
1916 4.4
1917 4.3
1918 0.6
1919 3.2
1920 0.2
Cravath's 1903-1906 seasons were not, in my opinion, good enough to get him promoted to the majors. I supposed that you could credit him with 1 WAR accumulated over various short major league stints in those years; however, he probably would have spent the bulk of those years in AAA.

Cravath now has 51.9 WAR; his top season is still 7.2 WAR, his top 3 are 18.2 WAR, and his top 5 consecutive are 24.8 WAR. Despite substantial minor league credit, Cravath is still below players like John Olerud, Cesar Cedeno, Bobby Bonds and Willie Davis. He is comparable to Robin Ventura, Phil Rizzuto, and Frank Chance; he's probably slightly ahead of Jack Clark. Ventura is 17th; Rizzuto is 26th (now that I look at my list again I feel that these two should be much closer together). Therefore, Cravath, with MLE's, is probably somewhere around 20-25 in my rankings.

Rating the 1880's pitchers:
Since I rated the 1880's pitchers much higher than the consensus, I was concerned that Sean Smith's WAR overrates 1880's pitchers; in particular, I was concerned that his replacement levels for the American Association and Union Association might be too high. So I decided to replicate his WAR for AA stars Mullane and King, but with different replacement levels.
I derived my replacement levels from the BP translation data and JeffM's estimated discounts, summarized in post 221 of the 1928 ballot thread. For 1882-1885, I used JeffM's data; for 1886-1888 I used an average of the BP data* and JeffM's data; for 1889-1890 I used JeffM's data, and for 1891 I used an average of the original BP data and Jeff M's data. For 1882-1885, BP and JeffM are almost the same; for 1886-1888, I didn't believe that the AA was actually equal to the NL so I felt an average was best; for 1889-1890, BP seemed too high, and for 1891, BP seemed too high and JeffM seemed too low. The average of 17.5% seemed right, given that in 1891, many AA stars returned to the AA from the Players' League.
* The original BP data had to be divided in half to account for an update to WARP; for 1886-1888 I used this updated, divided-in-half data, and for 1891 I used the original BP data.

Next, I multiplied Chone's baseline replacement level of .420 by each of these yearly percentages. I obtained the following replacement levels:
Year NL AA UA PL
1882 .420 .491
1883 .420 .466
1884 .420 .445 .546
1885 .420 .441
1886 .420 .428
1887 .420 .428
1888 .420 .426
1889 .420 .441
1890 .441 .487 .420
1891 .420 .494

Do these replacement levels look about right? What replacement levels does Dan R use?

Using these replacement levels, I came up with 62.5 WAR for Mullane and 58.4 WAR for King. However, I did not adjust for quality of opponents faced; such an adjustment would reduce King's total significantly since he played for the St. Louis Browns, the best team in the AA.
Anyway, if those replacement levels are correct, my initial rating of the 1880's pitchers was about right.

Never mind. John just answered my question (about player threads).

Dan R doesn't have data pre 1893.

Nice to see you back, jscmeagol!

BTW, meant to say welcome aboard Alex!

Ditto.

Re: Alex King #158

I join the others in saying welcome. Also, it's nice to know that someone still finds my Gavvy Cravath MLEs useful after all these years.

With respect to your conclusion that Cravath's walk rates are "unreasonably high," at first glance your comment seems reasonable, so I opened up my spreadsheet from five years ago and tried to figure how I came up with my estimates. First, though, I'll emphasize that we don't have any direct data Cravath's minor league walk rates, it really comes down what assumptions you're willing to make in extrapolating his major league numbers.

My estimates of Cravath's minor league walk rates started with his major league walk rates for the seasons 1908-09 and 1912-14. His average walk rate over these five seasons was .119. Actually, I split it in two parts, 1908-09 in the AL (.147) and 1912-14 in the NL (.112), and gave equal weight to each league, which gave me .129; but if I were doing it over again I would use .119.

Next, since I was expressing Cravath's MLEs in terms of the major league context of the season being translated, I adjusted for changes in major league environment. My assumption is that if walks became more common throughout the league, Cravath's rates would have changed proportionally. Taking 1911, for example, the NL walk rate was about 16 percent higher than it would be during 1912-14, and the AL walk rate was about 21 percent higher in 1911 than it had been in 1908-09. I also adjusted for aging using data from Tango's website, but this had only a small impact for 1910-11.

So looking back, there are three factors that cause my estimates for 1910-11 to seem "high": 1) I include his 1909 AL data (which you omitted from your table) when Cravath had a walk rate of .275, albeit in only 69 PAs; 2) I adjusted his rates to assume that he would have shared in the higher major league walk rates that were characteristic of 1910-11; and 3) I equally weighted 1908-09 and 1912-14. In retrospect, I don't think the equal weighting was a good decision, so if I were doing his MLEs again I'd probably lower his walks by about 8 percent. But dropping them all the way to .100 seems excessive to me.

On the other hand, it may not be enough to change his ranking. I switched to using Sean Smith's rWAR for my last HoM ballot, and Cravath just missed my ballot with minor league credit.

One early player I'll suggest you take another look at is Hugh Duffy, who has 49.6 rWAR before adjusting for season length. You also should keep in mind that season schedules were quite a bit shorter then--140 games from 1888-91, 154 games in 1892, 132 games from 1893-97, 154 in 1898-99, and 140 in 1900-03.

Brent/164:

In fact, I'm in the process of adjusting for season length and I have significantly raised my opinion of Duffy's career--he is now comfortably in my top 10, as the season-length adjustment pushed him to 59.3 WAR. Another player who benefited significantly from this adjustment was Ned Williamson, who I now have at 61.2 WAR. Chone really likes Williamson's glove, putting him at 91 runs above average in just over 5000 PA (when you correct for season length, he comes out at +150). Thanks for the advice!

Back to Cravath. I applied an 8% discount to your walk rates, and I recalculated Cravath's WAR for his PCL and AA years. The results:
Year New Old
1907 5.0 5.0
1909 3.7 3.5
1910 4.9 4.0
1911 7.9 6.2

Then I compared his walk rate (discounted from your MLEs by 8%) to league average (hit by pitches removed):
Year BB% above league average
1907 0.040
1908 0.056
1909 0.048*
1910 0.054
1911 0.059
1912 0.010
1913 0.016
1914 0.059
1915 0.068

*I used his MLE 1909 instead of his actual 1909 because the sample size on his AL performance is too small for it to be useful alone (I don't object to your adding it to his 1908 and 1912-1914 performance).

The abrupt discontinuity between 1911 and 1912 makes me somewhat skeptical of the 1909 to 1911 MLE walk rates. I think a downward trend would be more appropriate over these years--maybe from 0.050 in 1909 to 0.040 in 1910 to 0.030 in 1912. What if, rather than weighting each year equally, we applied a discount based on its distance in time from the year under consideration? What if we did something like a Marcel projection, but using Cravath's walk rate both before and after the year under consideration?

Brent/164:

In fact, I'm in the process of adjusting for season length and I have significantly raised my opinion of Duffy's career--he is now comfortably in my top 10, as the season-length adjustment pushed him to 59.3 WAR. Another player who benefited significantly from this adjustment was Ned Williamson, who I now have at 61.2 WAR. Chone really likes Williamson's glove, putting him at 91 runs above average in just over 5000 PA (when you correct for season length, he comes out at +150). Thanks for the advice!

Back to Cravath. I applied an 8% discount to your walk rates, and I recalculated Cravath's WAR for his PCL and AA years. The results:
Year New Old
1907 5.0 5.0
1909 3.7 3.5
1910 4.9 4.0
1911 7.9 6.2

Then I compared his walk rate (discounted from your MLEs by 8%) to league average (hit by pitches removed):
Year BB% above league average
1907 0.040
1908 0.056
1909 0.048*
1910 0.054
1911 0.059
1912 0.010
1913 0.016
1914 0.059
1915 0.068

*I used his MLE 1909 instead of his actual 1909 because the sample size on his AL performance is too small for it to be useful alone (I don't object to your adding it to his 1908 and 1912-1914 performance).

The abrupt discontinuity between 1911 and 1912 makes me somewhat skeptical of the 1909 to 1911 MLE walk rates. I think a downward trend would be more appropriate over these years--maybe from 0.050 in 1909 to 0.040 in 1910 to 0.030 in 1912. What if, rather than weighting each year equally, we applied a discount based on its distance in time from the year under consideration? What if we did something like a Marcel projection, but using Cravath's walk rate both before and after the year under consideration?

I did a Marcel projection for 1909, 1910 and 1911 using Cravath's 1908, 1909, 1912, 1913, and 1914 seasons (I left out aging).
For 1909, I weighted 1908 as 5, 1912 as 3, and his 77 PAs in the majors as 8 (this was an arbitrary decision by me and the weight may be way too high). I got a projected BB rate above average of 0.059, which becomes an actual walk rate of 0.131.
For 1910, I weighted 1908 as 4, 1909 as 5, 1912 as 4, and 1913 as 3. The projected BB rate above average was 0.029, which corresponds to an actual walk rate of 0.110.
For 1911, I weighted 1908 as 3, 1909 as 4, 1912 as 5, 1913, as 4, and 1914 as 3. The projected BB rate above average was 0.023, which corresponds to a walk rate of 0.107.
In a table, 1908-1915, but with the Marcel projections:
1908 0.056
1909 0.059
1910 0.029
1911 0.023
1912 0.010
1913 0.016
1914 0.059
1915 0.068

And actual walk rates:
1908 0.118
1909 0.131
1910 0.110
1911 0.107
1912 0.094
1913 0.093
1914 0.138
1915 0.140

What do you guys think?

I think we're splitting hairs and dealing with false precision. I admire the work though, this is more thorough than I typically get with MLEs.

Duplicate

Given the list at the header, I'd go Bagwell (I'm sure I'll vote for him), Walker (borderline depending on small things) and Brown (I have to take a look at the relative values of pitchers vs. position players for this era, which is a subset of the massive analysis I'm working on for 1880s pitchers). I doubt anyone else. Palmeiro's career makes me uncomfortable, because it bears no resemblance to a normal career, and there are no obvious ballpark effects or era effects to force that. Until I figure out what's going on, he's not on the list. Olerud and Franco would have to come up with a big surprise that supports them, but they're close enough for that to do the job. Other than that, I don't see anyone. - Brock

BTW, fra paolo - In case you happen to see this, you were kind enough to put up a link to a version of Bill James' old BROCK2 system, which I need to redo my Mark McGwire analysis. Unfortunately, I have a Mac, and all the links I can find either go to C code (not usable on a Mac) or a dead link promising the spreadsheet I need but not delivering. Do you have anything that gets the spreadsheet and is current?

Thanks, Brock

DL/168:

Here is Cravath's WAR 1909-1911. First, his WAR from Brent's original MLE's; next, his WAR from Brent's MLEs discounted by 8%; and last, his WAR using the Marcel prjoection.
Year BrentWAR -8% MarcelWAR
1909 4.0 3.8 4.0
1910 5.2 4.9 4.3
1911 8.2 7.9 7.1

I don't think a 2 WAR difference is "splitting hairs;" it could make a difference of 5 spots in my rankings (along with the reductions to peak value--that is, by using MarcelWAR instead of BrentWAR, Cravath's peak is reduced as well as his career totals).

BTW, fra paolo - In case you happen to see this, you were kind enough to put up a link to a version of Bill James' old BROCK2 system, which I need to redo my Mark McGwire analysis. Unfortunately, I have a Mac, and all the links I can find either go to C code (not usable on a Mac) or a dead link promising the spreadsheet I need but not delivering. Do you have anything that gets the spreadsheet and is current?

Brock, I have a zipped file on my PC I could send you. E-mail me via my BTF account page.

Alex,

On Cravath, I think we've reached the point where the differences between us are more art than science. You prefer to give more weight to his 1912-13 seasons because they come immediately after the period covered by the MLEs. On the other hand, I tend to see his walk rates during 1912-13 as a bit anomalous in view of his higher walk rates in 1908-09 and 1914-20. Intuitively, I find it appealing to imagine that his peak season came at age 30 rather than 34.

I have no problem with you making different assumptions than I have. I think we're all enriched when different voters explore different reasonable assumptions, and I think most voters have individually made judgmental tweaks to the various Negro league and minor league MLEs that have been presented here.

I'll emphasize that Cravath's walk rates are not the only area of uncertainty associated with these MLEs. I'm probably more concerned about the league quality and park adjustments. For league quality, I basically just assumed that the American Association of 1909-11 had the same relationship to the majors that Class Triple AAA leagues had 80 years later. And for park factors, I didn't have any direct data, but rigged up some very crude adjustments that are documented on the thread. If someone is willing to do the work, it should be possible to get better estimates in both areas, but it would require someone willing to put in quite a few hours of research.

Brent,

That's certainly reasonable about Cravath's walk rates, and I agree that it's good for voters to make individual tweaks and changes. This is especially important for someone like Cravath: with a large amount of uncertainty in his career totals, Cravath could easily be judged as a top 5 candidate or a #50 candidate. As it is, I am not totally comfortable with ignoring Cravath's 1914 walk rate when I calculate his walk rate for 1910; I also think that I should apply an aging adjustment to the Marcel projection.

If I get some time later in the week I might try to tackle the problem of the quality of the 1909-1911 AA.

Also sorry about the double post #166.

Recently I have been refining my rating system, adding adjustments for season length, discounting AA seasons, and considering two new categories: seasons with more than 2 WAR and seasons with more than 5 WAR. As a result, I believe that my rating system is far better than it was before. However, one very anomalous result remains: Silver King is still at ~16th in my rankings, and has become my highest ranked 1880’s pitcher. Extraordinary claims require extraordinary evidence, so in this post I will justify King’s rating, first by comparing him to his contemporaries and then by comparing him to pitchers I have ranked above and below him.

Silver King broke into the majors in 1886 as an 18 year old, pitching for the Kansas City Cowboys of the National League. In 1887, he joined Charlie Comiskey’s AA powerhouse St. Louis Browns, winning 32 games with a 120 ERA+ (which ranked 6th in the league). King had a monster year in 1888—a league-leading K/BB ratio, 585 IP, and a 199 ERA+. If we only allow seasons with at least 400 IP (a fairly typical workload for a pitcher of this era), King posted the second-best ERA+ of the 1871-1892 era, trailing only Hoss Radbourne in his incredible 1884 season. Furthermore, King was not dominating a dramatically inferior league; the AA of 1887 to 1889, though not quite as good as the NL, was very close, and far better than the AA of 1882 to 1885. And King wasn’t just the ace of an average AA team; he was the ace of the St. Louis Browns, who consistently dominated the American Association. After a decent 1889, King jumped to the Chicago Pirates of the Player’s League in 1890. King had another big year in 1890, leading the PL in ERA+ and finishing second in IP with 461. This season wasn’t quite as historically dominating as 1888, but King still managed to lead the PL in ERA+ by 28 points (162 to 134). After 1890, King rapidly declined; average seasons in 1891 and 1892 were followed by a terrible one in 1893, and King failed to impress in a comeback attempt in 1896 and 1897.

In many ways, King’s career path is similar to that of Dwight Gooden—both broke into the majors with outstanding seasons as teenagers, and both had historically dominant years as 20-year olds, before fading in their later 20s and washing out of the majors (although Gooden did hang on until he was 35). However, King had two top-flight seasons, while Gooden only had one; furthermore, King pitched more innings with a better ERA+ than Gooden. My ratings reflect King’s superiority, as I have King ranked 7th among all pitchers, while Gooden is 12th.

Comparing King to his Contemporaries

The top 6 pre-1892 pitchers, in my rankings, are (in no particular order), Charlie Buffinton, Silver King, Jim McCormick, Mickey Welch, Tommy Bond, and Tony Mullane. All 6 have roughly the same career value in baseballprojection.com WAR (when I apply my fairly harsh AA discount); McCormick stands out a bit with 59.1 career WAR (2nd place is Welch, with 55.4), while Bond is last, at 53.3*.

Pitcher TOT
McCormick 59.1
Welch 55.4
King 54.5
Buffinton 53.9
Mullane 53.4
Bond 53.3

*Includes estimated WAR for his 2 NA seasons: 0.7 for 1874 and 12.2 for 1875.

If I look at peak value, however, these pitchers are dramatically different.

Pitcher TOP3 TOP5
King 40.4 50.2
Buffinton 36.7 46.0
Bond 36.3 54.5
Welch 30.3 41.6
McCormick 28.3 43.0
Mullane 21.0 32.5

As a result, the 5-way tie obtained by merely looking at career totals (the differences between Bond and Welch are negligible) is resolved in favor of King, with Bond second and Buffinton third. Moreover, because of McCormick’s exceptionally low peak, I am inclined to rate him below these 3, although above Welch, due to McCormick’s greater career value. Mullane finishes last due to an underwhelming peak. Thus, I judge Silver King as the best 1871-1892 pitcher still on the ballot.

Comparing King to 20th century pitchers

As a peak candidate with a WAR total in the low 50s, King is most directly comparable to Kevin Appier, Vic Willis, Dwight Gooden, and Eddie Cicotte. In order to make King’s peak totals comparable to those of 20th century pitchers, I adjusted King’s seasonal totals by a factor equal to the three year average of the majors’ IP leader divided by 369, the average innings leader for the 1893-1923 era. Thus, King’s adjusted top-season mark of 10.7 is second to Dwight Gooden among that group of five, and he leads the others in top 3 seasons value while finishing second in top 5 seasons value.

Pitcher TOT TOP TOP3 TOP5
King 54.5 10.7 27.1 34.0
Appier 52.3 8.4 22.3 33.7
Willis 50.4 8.5 22.3 34.2
Gooden 49.4 12.5 22.3 30.1
Cicotte 49.5 9.8 24.1 32.2

Since King also leads this group in career WAR by a fair margin, I believe that he should be the top rated pitcher amongst this group. However, I am reluctant to rate King above Luis Tiant, who is the lowest-rated of my top 6 pitchers. Tiant has a sizeable edge in career WAR, 59.6 to 54.5. While Tiant’s top season and top 3 seasons are well below King’s (7.5 and 19.8 respectively), his top 5 seasons (30.6) are competitive with King’s. Tiant also has 6 seasons with more than 5 WAR; King has only 4. Lastly, Tiant has a remarkable 14 seasons with more than 2 WAR, which is tied for first among all pitchers, along with Brown and Reuschel. Such a measurement is not applicable to 19th century pitchers due to their far shorter careers; nevertheless, Tiant’s impressive showing is a definite point in his favor. As a result, King is well behind Tiant.

Sorry for the extremely long post; however, as a newbie, I feel that I must adequately explain my anomalous ranking of King.

So you are not his great-great grandson, Aluminum King?
sigh.

Thanks for this. The shattering of false hopes may be valuable.

Nope, no relation to Silver King.

In post 175, I forgot to explain my method for translating the AA pitchers King and Mullane. Using Brent's OBP+ and SLG+ factors (see link above in post 175), I calculated a wOBA+ factor, which I then multiplied by the NL's wOBA for that year, to get an estimate of what the average AA wOBA would have been in the NL. Then I calculated the runs difference between this AA average and the NL average using the approximation ((AA_wOBA - NL_wOBA)/1.15)*PA, where PA is the estimated number of plate appearances against the pitcher. Finally, I added this difference to the number of runs a replacement level pitcher would have given up, and moved replacement level to this new standard. As a result, 1882 was severely discounted (rep. level near .600) while 1883-1885 were strongly discounted (rep. level near .500); in 1886-1889, the AA replacement level was only a little higher than the NL replacement level.

You have no explicit treatment of errors or unearned runs, right?
ie, there is only whatever Brent has incorporated in adjustments to OB and SL averages

Paul:

I'm only considering OBP and SLG for the league adjustments; I'm ignoring errors and unearned runs.

During this down time here, I invite any HoMers looking for some ranking to do to check out The Seven Circles of Fame at Baseball Fever's hall of fame forum.

Do we still need an MVP project?

http://www.baseball-reference.com/leaders/WAR_leagues.shtml

Do we still need an MVP project?

No more than we need to ever vote for the HoM. Let's just line 'em up and put 'em in. Here's the next ballot for us all:

1) 79.9 Jeff Bagwell
2) 69.6 Tony Mullane
3) 67.1 Larry Walker
4) 65.7 Rafael Palmeiro
5) 65.6 Rick Reuschel
6) 63.9 Kevin Brown
7) 63.9 Jim McCormick
8) 60.7 Buddy Bell
9) 60.5 Sal Bando
10) 59.4 Luis Tiant
11) 57.4 Charlie Buffinton
12) 57.1 Willie Davis
13) 56.9 Bobby Bonds
14) 56.9 David Cone
15) 56.9 Tommy John

I was kidding. Baseball Reference just keeps getting better though. Now there's player photos for pre-1960 players. The big add I'd like to see is Negro League stats.

While that sounds pretty daunting, there are lesser things than the stats that would still have value. I'd like to see skeletal pages for Negro League players that show biographical information, years active and teams played for, predominant positions, and HoF and HoM status. (OK, even something as simple as place and date of birth might lead you off into the weeds of conflicting data, and "predominant positions" may not be all that easy). But even that much would be a very nice thing to have. And in the "awards" section, put in things like Pittsburgh Courier polls.

Another thing Sean could do is to get NPB stats - I assume that someone has those, and they would be in much better shape as data.

I'm pretty sure Sean hinted a few months ago that he was putting together some Negro League stats. There's still two big reveals to come in the next few weeks, so we'll see....

Not an HoM person, not nearly dedicated enough to go through all of the historical records, but I do see something that pokes out to me as a slightly "off" line of logic in Alex King's post about Silver King. Specifically:

And King wasn’t just the ace of an average AA team; he was the ace of the St. Louis Browns, who consistently dominated the American Association.

It seems counterintuitive to raise this as a positive point in King's favor almost immediately after addressing the issue of league competitive balance. If King pitched for the best team in the league, then his quality of competition was probably substantially lower (I don't know about the offensive levels of the various teams in the AA so perhaps the Browns weren't one of the better offensive clubs and my line of reasoning is flawed) and thus his performance is less impressive. Isn't this the same argument that people use against other players on teams that were traditionally dominant in otherwise weak leagues, like the 1950s(?) Yankees?

I am probably 100% wrong but would like to at least get this nugget of thought out there.

Do we still need an MVP project?

MVP Project.

Oblivion.

Your choice.

Daunte/186:

No, you're 100% right. The Browns had an above-average offense all three years King pitched for them, including an 1887 R/G mark of 8.20, compared to a league average of 6.58. Now that I re-read that sentence, I'm not sure why I wrote it--I might have been trying to say something like "King helped his team dominate the AA" (although using that as a point in his favor is also somewhat questionable). Anyway, when I calculated his WAR (imitating the methods of Sean Smith), I did include an adjustment for the quality of opposing offenses, adjusting replacement level to reflect the fact that King did not have to face the Browns. It's also worth noting that in 1890, pitching for a below-average offensive team in the stronger Players' League, King still managed to lead the league in WAR.

Perhaps we could see discussion threads put up for more of the top new candidates for the 2011 election? Kevin Brown, John Olerud and John Franco seem worthy.

I agree. Discussion threads for the new candidates are due. I don't see one for Larry Walker or Rafael Palmeiro either.

I'm going to ping everyone again after searching through the archives. Where is the in-depth discussion of Leon Day? I can't find his individual thread or much in the 1960 discussion. I'm embarrassed, especially after visiting the Negro Leagues Baseball Museum, that I've never formally analyzed his case. He seems like a Wes Ferrell or Bob Caruthers case to me. There's also war credit to consider. He made 7 all-star teams, surely we've given him adequate discussion.

I don't see one for Larry Walker or Rafael Palmeiro

Larry Walker

Rafael Palmeiro

If I could re-animate the discussion, I'd like to see where people rank Dick Redding, Hilton Smith and Leon Day. The comments I have dug up reveal a very incomplete discussion on Leon Day in particular. He looks like a "prime" candidate to me and he had a good enough bat to compile a lot of time in the outfield for various teams. He also lost time to the war so there are war credit issues. Redding and Smith are both close to my ballot so I'd like to see a good discussion. Unfortunately I don't have access to the year-by-year statistical record for Day to make a good comparison. Superficially he looks like Hilton Smith with a better bat.

Crossposting:

sunnyday2 Posted: August 30, 2005 at 09:21 AM (#1583937)
Eyeballing the various lists above, here is a plausible list of NeL candidates, in current order. One could of course quibble about the order, but I am absolutely sure I've got the right 7 pitchers 1-7, anyway, and all are for sure among the 30-33-35 pitchers in my hot 100.

The first tier has been elected.

Second Tier

1. Jose Mendez--ok, the lists don't support this at all. Mendez played too early for there to be much of a statistical record. He's no Smokey Joe, but who is? Not anybody else on this list.

2. Dick Redding--looks good on Black and Gray Ink, and All-Star and Best Pitcher selections--i.e. had a nice peak, and also a long career.

3. Hilton Smith--equals Redding on peak without the long career (though as long as many). We probably underrated him because who did we compare him to? Satch Paige.

4. Bill Byrd--does very well on BI and GI, wins, decisions and WAT, but spaced it all out a bit--i.e. had the long career that many others didn't have, but not the great peak.

5. Andy Cooper--up there on W and WAT, and accumulated them in a fairly short time--i.e. nice peak.

6. Leon Day--does well on All-Star and MVP picks, and GI, though he is the one guy on this list I don't feel like I've "finished" with. He is not unlike Hilton Smith his career curve, but without the BI. Could move up.

7. Nip Winters--this is probably too low for a guy who makes the All-Star and MVP list as well as the all-time W and WAT list. Had one of the greatest seasons ever. Paired with Andy Cooper.

From Leon Day's HOF info:

Leon Day
Born: October 30, 1916, Alexandria, Virginia
Died: March 13, 1995, Baltimore, Maryland
Bats: Right
Throws: Right
Played For: Baltimore Black Sox (1934), Brooklyn Eagles (1935), Newark Eagles (1936-1939, 1941-1943, 1946), Baltimore Elite Giants (1949-1950)
Elected to the Hall of Fame by Veterans Committee: 1995

Biography: The Negro leagues' outstanding strikeout pitcher with a dominating fastball and wicked curve, quiet Leon Day was the mainstay of the Newark pitching staff in the late 1930s and '40s. Also a superb contact hitter and speedy baserunner, Day was versatile enough to play second base or the outfield when he wasn't pitching. He spent two years pitching on integrated Army teams during World War II, and in his first game back with the Eagles in 1946, he tossed a no-hitter against the Philadelphia Stars.
Did You Know: that on July 23, 1942, Newark Eagles pitcher Leon Day struck out 18 Baltimore Elite Giants to set a Negro National League record?

Filling in the gaps, he had a nice 40 innings in Toronto 1951 age 34 (which makes his NGL debut at age 18).

He was in the war 1944-1945. I've seen reports that he played baseball and participated in D-Day.

1940 was an "outlaw" year in Mexico and Venezuela. He was quite good in Mexico that year.

I ran the post-1920 data from another thread through the filter I typically like to use for minor league data without run-scoring context which is K-BB/IP*9. This is a pure FIP type analysis but it generally gives me an idea of a pitcher's stuff. Here's my results:


Name IP RA K BB WHIP K/9 K-BB/9
Paige, Satchel 1506.7 3.31 1231 253 0.95 7.35 5.84
Smith, Hilton 812.3 3.37 470 96 0.95 5.21 4.14
Williams, Joe 565.7 4.50 266 70 1.12 4.23 3.12
Rogan, Bullet 1444.3 3.66 855 361 1.14 5.33 3.08
Donaldson, John 109 4.13 58 28 1.20 4.79 2.48
DiHigo, Martin 354 3.99 176 80 1.12 4.47 2.44
Foster, Willie 1859.7 2.87 1013 516 1.09 4.90 2.41
Day, Leon 513 4.53 288 154 1.16 5.05 2.35
Mendez, Jose 337.7 4.50 147 59 1.18 3.92 2.35
Byrd, Bill 1369 4.43 528 211 1.16 3.47 2.08
Cooper, Andy 1592.7 4.38 578 244 1.14 3.27 1.89
Brewer, Chet 1364.7 4.04 614 362 1.21 4.05 1.66
Bell, Cool Papa 293.3 5.65 134 84 1.34 4.11 1.53
Bell, William 1514.3 3.86 543 306 1.15 3.23 1.41
Brown, Ray 1284.3 4.16 445 254 1.17 3.12 1.34
Redding, Dick 628.3 4.71 217 134 1.33 3.11 1.19


This shows some interesting results.

1) Hilton Smith could really bring it.
2) Joe Williams was still awesome after 1920 despite his W/L record
3) Leon Day has stuff equivalent to Dihigo, Bill Foster and late era Jose Mendez - 3 guys we elected
4) We've rightfully ignored Andy Cooper
5) Is there missing data for Ray Brown? His numbers look awful.
6) Dick Redding didn't add much value to his resume after 1920.

On Leon Day, ultimately I've decided to slot him in around #35 on my ballot. I don't like him nearly as much as Hilton Smith. I've decided I like Smith more than Dick Redding also. Leon Day does have a good case though and is still rather sketchy - I don't have the year by year stats available to do a proper WAR analysis though I'd love to see one. Bucky Walters supporters might be able to get behind another pitcher who could hit and had a nice 7 year prime.

Leon Day WAR Analysis

A few months ago, I computed some rough WAR estimates for Day based on Chris Cobb's MLEs. Rather than just using Chris' DERA estimates, I regressed them, as the straight DERA estimates showed wild swings in performance. I calculated regressed DERA by averaging Day's actual DERA and his average DERA for the 3 surrounding years. From these DERAs, I calculated pythagenpat win percentage and then WAR, using a replacement level of .420 WP. In addition, I estimated Day's war years at 4.0 WAR each; this is nothing more than a wild guess, based on my assumption that Day's 1935-1943 peak would have continued through the war years.

Year WAR
1935 4.3
1936 4.6
1937 8.3
1938 0.7
1939 3.3
1940 4.9
1941 5.8
1942 4.5
1943 1.1
1944 4.0
1945 4.0
1946 2.1
1947 0.6
1948 0.9
1949 0.3
TOTAL 49.5

I think that Chris' MLEs exclude Day's minor league numbers from 1951 to 1953 (his age 34-36 seasons). Based on Day's outstanding 40 IP in the IL in 1951, he may deserve additional MLE credit for 1950 and 1951 (Day played in the NEL in 1950, but according to Chris Cobb, there is no data for 1950).

In my updated rankings, I have Day at #85 (based on the WAR estimate posted above). For comparison, I have Redding at 9, Smith at 10, Matlock at 44, Byrd at 100, Winters at 103, and A. Cooper at 120.

How much of that is hitting? Day played 2B/CF when not pitching because he was just good enough with the bat.