Numbers on Willis

First his actual numbers
249-205, 2.63 ERA, 118 ERA+, 3996 IP

He seems to have gotten average run support, according to RSI (100.13), making his record 250-200.

According to BP's stats....
...His 4.13 DERA puts him 9th behind (in no particular order) Rube Waddell, Addie Joss, Urban Shocker, Eddie Cicotte, Dazzy Vance, Clark Griffith, and Eddie Rommel. Sam Leever is at 4.14. In other words, pretty good considering we have 25-30 pitchers in our consideration set, but not anything of note.

...his tranlated record is 164-161 (anyone know how they calculate this, it seems screwy for some players) and they chop off roughly 1000 IP, down to 2922.7.

His K# is 113, with 6.1 translated K/9 according to BP.

His BB# is 109, with 3.2 BB/9 according to BP.

His K/BB# is 110.

His K# isn't terribly great but his tranlated K/9 is behind only Wadell, Vance, Bender, Marquard, Shocker, and Grimes. 7th out of 30 isn't bad but it isn't noteworthy either.

Overall, Willis looks good but doesn't jump out at me. I would have him above his exact comteporary Leever because he threw about 1000 innings more in the exact same time period, at a pace that was very comparable.

However, he hasn't cracked my top 50 which includes Vance, Waddell, Griffth, Rixey, Joss, Shocker, Mays, Cicotte, Welch, Grimes, Luque, and Rommel. This makes 14 pitchers when Redding and Mendez are counted. Willis is probably the top of the leftovers and I will take another look at him. On second thought, he may be better than Luque,Mays, and even Welch.

Leever's better -- MUCh better W/L record and better ERA+ and he only started late because of the screwy economics of the 90s NL

Oh, I should mention I have no real way to judge peak for pitchers yet besides eyeballing WARP and WS. Hope this gets the conversation going!

Karl,

All of those extra IP have to count for something right? And while Leever's record is better, his peripherals are not, so we are left with much a dilemma much like the one we face with Welch, just less so. Leever also got much better run support, a 112.92 RSI.

Leever may very well be better, it is tough with these pitchers as so many of them seem alike.

And I am not familiar with how Leever's career got started late, what are the details? This could color my perception of him.

I just don't get Karl's obsession with Leever.

He worked as a schoolteacher for at a time in his 20s. Karl has made the leap to assuming that he chose teaching over a baseball career because of the economics of the time. The actual evidence, as I recall, was that he stunk in his early 20s, and chose schoolteaching because he wasn't much of a ballplayer. While teaching, he developed a curveball, and thus returned to baseball.

I fail to see any rational basis for giving him extra credit.

How much did school teachers make at the time? I would assume it isnt' much. I know there was a maximum salary for NL players in the 1890's, 2400 for something.

Schoolteachers did better than scrub ballplayers in the 1890s. When you look at the details it's pretty clear that he only changed to baseball when the economics staretd improving, after the slump of '93-96 ended. He had better alternatives than most ballplayers, being an educated man, so he took them. Once he started in ball, he took only 1 year in the minors before reaching the majors, then 1 year in the majors before becoming a star. Essentially his years 22-26 (say) were blocked by the MLB and national economics of 1893-97. In 1892-93, when he'd normally have started ball, leagues were going bust all over the place and the US economy was heading into a tailspin. Leever was blocked from MLB by the economy just as surely as by wars or by being the wrong color. This is why we have a player dearth from the 90s.

A star pitcher for several years on club teams in southwestern Ohio, Leever was a teammate of future major leaguer Kid Elberfield with the local Norwood Maroons. He taught school during the week and pitched on Sundays. A big strong right-hander with an exceptional curveball, Leever was not signed sooner likely because he was not a hard thrower-he relied more on control and the movement of his pitches.

From the Baseball Biography Project

Quoting from Kelly from SD in post 48 of the 1932 ballot thread:


Re: Sam Leever,
From SABR's Deadball Stars of the National League.

....


1889: graduates high school, begins working as a schoolteacher and pitching on Sundays for semipro teams in southwestern Ohio.
No professional interest because he "lacked an impressive fastball."
1896: Signed first pro contract after Pirates discovered him when he was playing a game called "Anthony Over," a game where people toss a ball over a barn to one another and he supposedly curved the ball around the barn.


So basically, he was not playing baseball for a living because the pros didn't want him. When they decided they did want him, he became a ballplayer.

I still see zero evidence that he chose an alternate path for economic reasons, other than pure speculation. And even if there was such evidence, I don't see how he could be given extra baseball player credit for making a choice to not play baseball. Either he wasn't considered good enough or he chose to do something else. Neither of these are reasons for extra credit.

Either he wasn't considered good enough or he chose to do something else.

Could you imagine how many careers we could extrapolate based on that? We would have two thousand players in the HoM if we went that route!

Re: Willis

What helps him (and Waddell for that matter) is that pitchers who started at the beginning of the 20th century didn't have very long careers on average, so I give him a little boost for that reason. Of course, his peak also has to be normalized.

Leever clearly developed something between 1889 and 1896. How aggressive was scouting in 1891-96, when teams were going bust all over? I would guess, not very.

Leever's career is what it is, it's not 4000 IP. At the same time, while one normally discounts a shorter career pitcher compared to a long career pitcher because the extra IP would have come in a decline phase, that's not true for Leever, whose extra IP would have come in his mid 20s.

And his W/L is a LOT better than Willis, and as far as I can determine, his contemporary reputation was higher than Willis -- he was more "famous" (yes I know we're the HOM not the HOF but contemporary F is an important though not concluisve indicator of M.)

And his W/L is a LOT better than Willis

Leever's supporting cast was a LOT better, too.

RA+-Pythagorean records for Willis

      Raw data       Adjusted for defense
Year    W-L    FWP     W-L     FWP
1898   21-14    20    18-16     11
1899   28-10    38    24-14     26
1900   14-12    10    12-14      3
1901   24-10    29    21-13     22
1902   27-19    23    26-19     22
1903   18-13    16    19-12     19
1904   18-21     4    20-19     10
1905   17-21     4    17-21      5
1906   24-11    30    24-11     29
1907   19-13    17    19-14     16
1908   18-15    13    18-16     11
1909   20-12    21    18-14     15
1910   10-14     0    11-12      4

Total 258-186  222   248-196   192

As for Leever, I have his "raw data" RA+-Pyth. record at 179-117 (169), with a best year of 22-9 (29) in 1903. I haven't done an "adjusted for defense" for Leever, but it would be at least as important for him as for Willis.

My point of view: there's no way to think of Leever and Willis as any kind of equivalent; Willis belongs many places ahead of Leever on any ballot.

I've been running Willis at about 25th on my priority list; half the time I think that's right and the other half of the time I think I should have him well above that and on the ballot. Advocates of Griffith point out that our list of active HoM pitchers is unusually short in the late 1890's. Willis would be another candidate to fill that hole.

Pythagorean adjustment doesn't work in a period (before 1913) that didn't know what ERA was -- they weren't pitching to it. Leever's ACTUAL W/L are pretty damn good -- so what if he stayed in a few blowouts and allowed more runs.

Pythagorean adjustment doesn't work in a period (before 1913) that didn't know what ERA was

Yeah, but they knew what runs scored were. I doubt any pitcher was pitching to the score early on during a game. He was trying to prevent any runs from scoring.

The biggest knock on Leever as compared to Willis in my mind is that Willis seems to have pitched many more innings in exactly the same time period. Many more decisions too. Too add to that, their ERA's are very close and their DERA's are even closer. So they pitched at the same level and Willis threw many more innings, getting many more decisions.

But ERA was known before 1913, even though it wasn't an official statistic. Similar to saves in the 1960's. It dates back at least to the 1880's. At that time, an earned run was considered to be one that scored only as a result of base hits.

Plus, do pitchers really pitch to ERA nowadays?

I just posted under/overachieving pitchers eligible for election around 1940/1 in the Rube Waddell thread. If anyone's curious, you can check there, or go to the link to that page on my site given there.

Free Pink Hawley! :)

I'm thinking of placing Vic Willis at 12th on this season's ballot (ahead of Redding and Bridges). Win Shares data would be helpful in determining if that is too high.

Vic Willis is a guy I tried to find a spot for on my ballot about a decade and a half ago but just couldn't do it. He currently resides at #29 on my ballot.

Best 10 years in descending order:

39,33,29,29,25,24,20,20,19,19,17

Career: 293

I don't know if this is schedule adjusted or not. But then again I may have decided not to schedule adjust WS for pre 1920 pitchers since they threw so many innings anyway.

Remember, however that WS will overrate pre 1920 pitchers because they threw a lot of innings per year. Defense was a bigger part of the game in that era then WS lets on as well.

I wouldn't put him above Redding (who is my #1 overall pitcher) but above Bridges sounds right. Then again, Bridges isn't really on my radar.

I'm one of Willis's best friends (had him 6th in 1968), and that peak looks pretty scrumptious to me, even if you discount a bit for the reasons jschmeagol mentions. I may be discounting a bit less than a lot of others, though--I'm not entirely convinced the factor is that great, and to me Willis's peak stacks up pretty well even against his own contemporaries.

Quite short peaks are enough for me, though, and most disagree with me on that.

Well, my research today bumped Waddell back onto the ballot ahead of all three.

>Remember, however that WS will overrate pre 1920 pitchers because they threw a lot of innings per year.

Say what?

Remember, however that WS will overrate pre 1920 pitchers because they threw a lot of innings per year.

As a general rule, pitching staffs get 1/3rd of the Win Shares in any given season. On a team basis, this goes up or down somewhat depending on whether the team has a good or a bad offense, or if the fielders have good or bad DER. But league-wide, about 1/3rd of the WS goes to the pitchers.

Under WS, 250 IP at 100 ERA+ will get about the same number of WS regardless of the baseball era. In 1879, that's the backup starter, understudy to the ace who tosses the other 500 IP. In 1920, that's a typical third starter. In 2005, that's more IP than anybody actually throws. Under WS, their value is approximately the same.

As the workload of an average starter decreases, so does the WS for an average starter. It's up to you to decide if an inning is always an inning, no matter when it occurred, or if innings have gotten more difficult (and more valuable) over the years.

Me, I think WS underrates modern pitching, and we are starting to see those effects as we evaluate 1950's pitchers.

Under WS, 250 IP at 100 ERA+ will get about the same number of WS regardless of the baseball era.

That is, about 15 WS.

>Remember, however that WS will overrate pre 1920 pitchers because they threw a lot of innings per year.

Say what?

Well, the assertion that WS will overrate pitchers simply because they threw a lot of innings is not correct, but the point that win shares will overrate pre-1920 pitchers is correct.

The win share formula for splitting credit between pitchers and fielders assumes that a 67%/33% pitching/fielding split of responsibility for defense. Under any conditions of play in which fielders actually have a higher share of defensive responsibility, win shares will underrate them and overrate pitchers.

This overrating becomes more extreme the farther back in time you go, but I think it is correct that all pre-lively-ball pitchers are at least slightly overrated by win shares.

Willis is overrated by win shares partly for this reason and partly because of another, related feature of the win shares system, which caps the percentage of a team's win shares that can be earned for fielding, even if the main win-share formula indicates that they deserved more. This happened quite frequently with great defensive teams before 1910: it's almost an annual thing for the great Cubs teams of the aughts and for the Braves through the 1890s. Willis's great early peak as win shares sees it gets quite a bit shaved off if full credit to the fielders is given.

It is matters like these that cause the electorate to decide that Willis was not a strong candidate 45-50 years ago. More recent arrivals might wish to mull over these matters in considering Willis's candidacy.

Even if one is skeptical of these arguments, it is worth considering how Willis compares to his peers. We've elected a pretty good number of pitchers from that era (Young, Mathewson, Plank, Walsh, Brown, McGinnity, Rube Foster) and Rube Waddell and Addie Joss also have cases to be next in line, though their cases depend on how you weight peak value and value per inning vs. career value.

I'm not a very big fan of Win Shares for pitchers. The pitching/fielding split isn't dynamic enough . . . and unlike hitters, context is much more era dependant (I'm talking IP/teamG here). I just try to look at as much as possible - big years, translated innings, ERA+, WARP, etc. and make a best guess.

I like Willis - I have him around 50 IIRC, and I wish I could push him higher. He's got some really big years in there, no bad ones, except for his last. You could certainly make the case he's better than Griffith.

He's got 676 more translated IP, which is almost 3 years worth. But both pitched for about 13 years, which really means Willis was much more durable in-season, or used more, or whatever - he pitched about 25% more innings, which is a lot, considering their careers were the same length. Willis ERA+ 118, Griffith 121. But Willis has a bigger ERA+ peak too, 3 seasons over 150, to 1 for Griffith, whose next best is 134.

The knocks, Willis got much more defensive support, NERA-DERA is -.15, Griffith is +.01. Griffith was a much better hitter, he even played the field a little bit. These two things give Griffith slightly higher WARP1 and WARP3 (80.7-78.1; 75.1-70.2) - but he should not get a bonus for achieving this in fewer innings, when compared to Willis. This is very important - their careers were the same length, so there wasn't any extra pennant impact. Griffith's career was Frank Chance shorter, not Albert Belle shorter. Very important distinction.

So if you think the defensive support and their own hitting matters, you nudge Griffith ahead. I should probably have them closer than I do. I have Griffith 9th, and Willis 52nd. That's probably wrong. I'm still not sure if it means Griffith moves down or Willis moves up. Probably a little of both.

Griffith's career is longer than Willis's if you give him credit for the 1892-93 seasons, which he lost to contraction. If the league doesn't shrink, Griffith picks up 400-500 more major-league innings pitched.

He also pitched more in short seasons than Willis did.

Also, if you use WARP to compare the two -- and I agree with Joe that WARP is more accurate for pitchers in this era than WS -- notice that Griffith won 11 more games than his stats would predict, while Willis lost 11 more. This is no knock on Willis: most of the good pitchers are in the negative here. But Griffith, who had a reputation for smarts, appears to have indeed beat his statistical projections. WARP doesn't give him credit for that in its comprehensive assessment (if I understand their system correctly), but I think it's a point in Griffith's favor.

Good points Chris . . . and I forgot about the 1892-93 thing.

Well, my research today bumped Waddell back onto the ballot ahead of all three.

No argument there! (My pitching order at the moment is Méndez-Waddell-Willis.)

I agree, by the way, that WS overrates Willis; if I didn't, he'd be by far my #1 pitcher. But Willis does pretty well (not AS well, but well) in other measures too, including PRAA, one of my favorites.

As for Griffith, I've definitely had him too low (my PRAA year-by-year peak analysis proved that, too). I'm moving him up yet again this year. That would help my consensus scores if it weren't for the fact that Cicotte may move up as well.

I also agree that WS is starting to underrate pitchers as we enter the 50s and 60s; however, I still need to see a strong peak, and many of the guys we've seen so far don't have one even after you adjust for this.

Other reasons to prefer Griffith over Willis include:

1) ERA+ is harder on Griffith, because it's harder to dominate a contracted league.

2) Willis played most of his career (after 1901) in the lesser of the two leagues.

3) We've elected many more pitchers whose careers centered in the aughts than in the 1890's.

2) Willis played most of his career (after 1901) in the lesser of the two leagues.

That's an arguable point. Dick Cramer's study from the seventies didn't see any real difference between the two leagues during that time, though it spots the differences between the leagues during the 1880's and 1950's, for example.

>I also agree that WS is starting to underrate pitchers as we enter the 50s and 60s;

To me this is a dangerous assertion stated baldly like this, just like the converse that WS overrates pitchers of the deadball era (because they threw a lot of innings).

I will agree that WS overrates pitchers pre-1893 because the defenses are playing a larger role. And if anyone wants to argue that defense continues to play more of a role through the deadball era than it does more recently because of Ks, I wouldn't argue that point too much.

But for the most part the differences in pitching WS in the 20th century is a function of IP. A team that won 100 games in 1900 maybe split most of the WS among 45 pitchers. In 2000 that same team split most of the WS among 7-8 pitchers or more. The average "contributing" pitcher in 1900 clearly has more value than the average "contributing" pitcher in 2000.

Mitigating that is factoring in where a pitcher stood among his peers, which is a way of asking not of value per se but of differential or marginal value versus other "contributing" pitchers or replacement value, etc.

I guess I'm just saying what has been said a thousand times--there's "value" and there's "marginal value," and the problem of what the baseline for marginal value is continues to be elusive.

Anyway I don't think WS underrates modern pitchers per se, I think it properly values individual pitchers in a historical context. It's just that raw WS are less important than marginal ones. It just a shame that we can't have more confidence in PRAR and PRAA because of the various machinations that are hidden within the numbers.

The point here is not that Willis and Leever are being undervalued (or overvalued), just that the generalization "WS overrates deadball pitchers" shouldn't be used to dismiss a whole cohort.

But in 1900 defense was a much larger part of the game than it is now. Because of this pitchers didn't have to throw as hard or as often to get outs so they could throw more innings. It is a change in the nature of the game that needs to be taken into account. WS doesn't track this affect, it keeps the pitching/defense balance static and then since pitchers form 1900 are throwing more innings, it gives them a ton of WS.

I don't know if this is schedule adjusted or not. But then again I may have decided not to schedule adjust WS for pre 1920 pitchers since they threw so many innings anyway.

Suppose so. There may be for any career length only one span in mlb history when the two biases balance. Even so, even concerning one pitcher whose career fits the time span, it is vital for anyone who pays attention to single-season measures to adjust for schedule length. Otherwise one will be fooled because Vic Willis put up his best two years, measured by rates, precisely here.

132 132 132 132 <u>154 154</u> 140 140 140 140 (games scheduled NL 1894-1903)

It is a saving grace for shallow comparison of these two pitchers in particular, Clark Griffith hit the same two years.

Regarding the careers, the 15-year average length of schedule, looking forward from the time Griffith arrived in the majors was 141.1 games; when Willis arrived, 150.3 games.
Length of Early MLB Seasons

Marc--

Good point of clarification. I really meant simply that using raw WS to compare guys like Spahn to guys like, I don't know, Waddell is starting to do the modern pitchers some injustice. You're right as to the cause of that injustice, and I think it's easy to overdo remedies for this.

But I do think that voters like me (OK, it's true, there aren't many) need to be careful we're not underrating 50s and 60s (and later) pitchers, compared to the earlier ones, as the WS numbers begin to drop. (Remember, I've had all these guys--Lemon, Wynn, et al.--a lot lower than most have, and the ML pitchers who are highest in my rankings include Waddell, Willis, Cicotte, and Griffith. A lot of that is simple peak vs. career preference, but I want to be sure there's not something else underlying it.)

For those of you looking to baseline WS between eras with a replacement level - try subtracting 6.5 WS for every 220 IP. I think that will help with the era adjusting.

I think the changing defensive impact not all that significant once you get past 1920. The pitchers are still being compared to other pitchers within their respective leagues and years (each league basically has a set # of pitching WS to distribute, the K and HR adjustments to the defense/pitching split are quite minor), so WS sort of stabilizes the changing defensive impact across eras.

IMO this makes it not great for describing the changes from era to era, or what is happening in a particular era. But, it helps rather than hurts cross era comparisons of pitchers. I say this as long as the basic game is the same - the basic game pre and post 1920 is much different, obviously.

try subtracting 6.5 WS for every 220 IP.

I believe that you meant "330 IP". A league average pitcher with 220 IP gets 13 Win Shares. 6.5 is half of his value. A league average full-time batter/fielder gets about 19 WS. 6.5 is one-third of his value. The adjustment as written kills pitchers.

The above assumes that WS replacement is the same for both offense and defense. Many agree that WS is imbalanced with a much lower replacement level on the offensive side of the equations. (Which IIRC is why tangotiger proposed a 39/61 offensive/defensive split, to balance the replacement levels.)

I think the changing defensive impact not all that significant once you get past 1920.

If this were true then pitchers today should be pitching about the same amount of innings as pitchers in the 1920s. I think that the value of top pitchers tends to remain fairly constant across eras, and that the IP levels have dropped because the individual innings have become more valuable (hence more difficult).

Average innings-pitched for a starting pitcher stays near to constant from 1925 to 1970. Here are the results of a study I have done, tracking the ip of the top 4 starting pitchers on each team, season by season, from 1911 to 1970, to find the average innings thrown by a major-league starting pitcher, decade by decade. I split the 1920s into two parts because pitcher usage was clearly in transition from 1921-25.

1911-20
252.3 ip
18.2% of team ip

1921-25
228.7 ip
16.5% of team ip

1926-30
218 ip
15.7% of team ip

1931-40
218 ip
15.7% of team ip

1941-50
206 ip
14.9% of team ip

1951-60 (+1961 NL)
201 ip
14.5% of team ip

1961-70
211 ip
14.5% of team ip

As you can see, average ip declines steeply during the early twenties, then levels off for 15 years, declines a little more, the levels off for another 20 years from 1951-70.

I haven't carried my study past 1970 yet, since we won't be considering players who were active after 1970 for 7 years yet, but I'm guessing that the totals will be pretty steady for that decade as well, though its possible that there will be a significant difference between the two halves of the decade, or between the two leagues after the addition of the DH.

I expect that the next significant decline will arrive in the 1980s.

Chris, did you also track BFP?

And could you show the the average runs per game and/or the average baserunners per game?

Just wondering what those contexts would show...

If pitcher IP dropped again in the '70s, what else does the '70s have in common with the '20s? Well, relatively speaking (and only from a relative standpoint) there was an increase in offense. A huge increase in the '20s and a slight increase in the '70s. Not enough of an increase, you wouldn't think, to "revolutionize" pitching. So I'm more inclined to think that there was something else in addition to the added offense. But what? I'm not sure. Maybe it was nothing more than a new mindset, a mindset that said, dammit, I am not gonna lose this game while I've got a fresh arm out in the bullpen.

IOW in the '70s maybe it was mind over matter whereas in the '20s it clearly was matter over mind. Just a hypothesis.

I've studied the same thing from a slightly different angle, the concept of a "median ace" which concentrates on the top N workhorses, where N is the number of teams in MLB.

My results agree with yours. A 5% decline from the late 20's to the late 50's. The 60's recover that 5% due to the schedule expansion (two extra starts on rotation didn't seem to overload anybody).

Looking forward, the early 70's have a rise (DH?) with totals like the early 20's, and then comes the decline, with the 1990's 22% less than the early 1970's.

A 5% decline

Typo alert. 8% decline.

Chris, did you also track BFP?

No, and I don't have access to the data to do so, unfortunately.

And could you show the the average runs per game and/or the average baserunners per game?

I don't have that information, but r/g at least would be fairly easy to compile.

Just wondering what those contexts would show...

It would be interesting. I suspect that BFP data would show a very gradual and very steady decline from 1925 to 1970, which is masked in the periods of IP stability by gradual declines in offensive production. IP hold steady while offense declines gradually from a peak in 1930 to WWII. When the offensive production jumps again in the late 1940s, IP drops. As offense gradually declines again to the late 1960s, IP hold steady.

Because the steady counts in IP are across periods in which offensive production gradually declined, I suspect BFP were gradually declining during those periods also even though IP were holding steady.

Chicago Tribune 1900-07-16, coverage of yesterday's game at Chicago (Chicago 5, Pittsburgh 3)

Higginsport is Avenged.
Jack Taylor gets even for an old defeat by Leever.
[. . . more subtitles]
_In 1894 the Martinsville (O.) team defeated the Higginsport (O.) team at Martinsville by a score of 4 to 3, and in celebration of that event all the residents along Dodson creek adjourned to the Lynchburg distillery.
_At the next meeting of the schoool directors of Goshen, O., Samuel Leever, who pitched the game, was voted an increase in salary as teacher, while his catcher, Jerry Kiskadden was made road supervisor. Jack Taylor pitched for Higginsport.
_Yesterday Taylor had his revenge.

As I said, Leever deserves credit for 1894-97, given MLB's insecurity, restricted opportunities and poor prospects in those years. He's a victim of the '93-96 recession and the one-league system.

As I said, Leever deserves credit for 1894-97, given MLB's insecurity, restricted opportunities and poor prospects in those years. He's a victim of the '93-96 recession and the one-league system.

Wish this was Willis and Wilbur, I always get those 2 mixed up.

Chris Cobb #28 and #30
. . . Willis is overrated by win shares partly for this reason and partly because of another, related feature of the win shares system, which caps the percentage of a team's win shares that can be earned for fielding, even if the main win-share formula indicates that they deserved more. This happened quite frequently with great defensive teams before 1910: it's almost an annual thing for the great Cubs teams of the aughts and for the Braves through the 1890s. Willis's great early peak as win shares sees it gets quite a bit shaved off if full credit to the fielders is given.

It is matters like these that cause the electorate to decide that Willis was not a strong candidate 45-50 years ago. More recent arrivals might wish to mull over these matters in considering Willis's candidacy.

30. Chris Cobb Posted: January 27, 2006 at 11:02 PM (#1841230)
Griffith's career is longer than Willis's if you give him credit for the 1892-93 seasons, which he lost to contraction. If the league doesn't shrink, Griffith picks up 400-500 more major-league innings pitched. He also pitched more in short seasons than Willis did.

In the eyes of anyone who glances at readily available statistics, including ERA+ and Win Shares, Vic Willis gains as much as any player from some good fortune.

His two great seasons at the end of the 1890s coincide with a two-year trial for the 154-game schedule. They got to 154 in 1898 by increase from 11 opponents at 12 games each (11@12) to 11@14. With contraction to eight teams two years later, they revised the schedule to 7@20, which had been the eight-team-league standard before the NL-AA merger.
For those two great seasons, Willis also played on a great team, repeat champions in 1898 and a serious threat to the syndicate Superbas in 1899. A few Boston players were renowned fielders and a few more seem in retrospect to be worthy of renown.

--
13. John (You Can Call Me Grandma) Murphy Posted: December 11, 2004 at 08:23 PM (#1013653)
> [Karlmagnus: Sam Leever's] W/L is a LOT better than Willis

Leever's supporting cast was a LOT better, too.

14. OCF Posted: December 12, 2004 at 01:20 AM (#1014483)
RA+-Pythagorean records for Willis

Raw data djusted for defense
Year W-L FWP W-L FWP
1898 21-14 20 18-16 11
1899 28-10 38 24-14 26
1900 14-12 10 12-14 3
1901 24-10 29 21-13 22
1902 27-19 23 26-19 22
1903 18-13 16 19-12 19
1904 18-21 4 20-19 10
1905 17-21 4 17-21 5
1906 24-11 30 24-11 29
1907 19-13 17 19-14 16
1908 18-15 13 18-16 11
1909 20-12 21 18-14 15
1910 10-14 0 11-12 4

Total 258-186 222 248-196 192

If I understand correctly Willis gains from OCF's adjustment for team defense in three seasons only, 1903-1904 and 1910, and he gains only four games total. That matches what he loses by the same adjustment in 1899 alone.

I believe that his play for horribly bad teams helped make Vic Willis for the Hall of Fame. But he played two seasons for one great team (Boston 1898-99) and four seasons for another (Pittsburgh 1906-09). And Boston remained a strong team through 1900, maybe a strong fielding team longer than that. Jimmy Collins, Chick Stahl, and Billy Sullivan moved to the American League. But Herman Long and Fred Tenney, two of the renowned fielders remained.
(Hugh Duffy is another renowned fielder who moved, but he was past the productive part of his playing career. Billy Hamilton and Bobby Lowe are two more regulars who remained, and one 'metrician rates them highly --Bill James, I think.)
(Two other regulars moved to the AL, total 6 of 10. I know nothing about catcher Bill Clarke as a fielder. Outfielder Buck "Home Run" Freeman was considered a poor fielder.)

Leever played for Pittsburgh only. Essentially that was a great team for ten seasons, 1900-1909, which were no longer full seasons for Leever at the end. It was an average team before and after, including the only season Leever pitched a boatload of innings, 379 as a rookie in 1899.

I'm confused here, Paul. Are you disagreeing with my assessment that Leever pitched for much better teams than Willis did?

I believe that his play for horribly bad teams helped make Vic Willis for the Hall of Fame.

Whatever the reason he made it for, I'm still glad that he did. :-)

I wrote in #53:
I believe that his play for horribly bad teams helped make Vic Willis for the Hall of Fame.

I mean that the Hall of Fame case for Vic Willis turned partly on his play for horribly bad teams. That is a distortion because he also played for great teams.

Probably I shouldn't have quoted the brief exchange between karlmagnus and you, John Murphy, because (a) it doesn't add much if anything, (b) it may be focused narrowly on W/L and (c) it's clear in retrospect that you shouted "a LOT better" only because he had done so. My comment focuses on fielding support, as intended, which is a poor fit for the W/L context that I provided.

Overall, not limited to fielding support, Willis enjoyed a great team in six of his 13 seasons. That's good enough I would say only one, possibly two pitchers in the major leagues played for "a LOT better" teams: Al Spalding. On the other hand, Leever did play for better teams --great teams in ten of his 12 seasons? Ignoring time over-the-hill, it is 6/12 for Willis and 9/10 for Leever.

--
Some context:
I believe I read/skimmed this entire thread in response to the Keltner List for Vic Willis by 'AG2004' (Spalding returns?) at Baseball Fever.
Presumably, I dumped the clippings into this composition window as I proceeded, which essentially means "maybe say something about this . . . maybe say something about this".

Thanks for the explanation, Paul.

I'm trying to put together an initial ballot for the 2016 election based primarily on my Player won-lost records. Player won-lost records are calculated based on Retrosheet play-by-play data, which don't extend back to any of Vic Willis's career, so I have to try to figure out how to fit him into my consideration set judgmentally.

I've mentioned in a couple of places that my system really loves starting pitchers, especially slightly above-average starting pitchers, which would seem to suggest that I should really like Vic Willis. But since I haven't applied my system to the Deadball Era, I'm a little leery of knowing exactly what might happen there. And I have no idea what my system would make of a pitcher throwing 410 innings in a single season.

Reading through this thread and looking at Willis's career, the more modern pitcher that he kind of reminded me of, in terms of career shape and quality, was perhaps Jim Bunning. And looking through the Jim Bunning discussion thread, lo and behold, at Comment #109 OCF presented a set of his PythPat W-L records which show Bunning with a career record of 238-180 and Willis at 248-196, with very similar top 5 seasons as well.

What do others think of the Willis-Bunning comparison?

Also (and I may repeat this in the 2016 Discussion Thread), I really like OCF's PythPat W-L records as an analysis tool for starting pitchers. OCF, is there any chance that you have a spreadsheet with these numbers for most pitchers that you'd be willing to share? Or even a set of formulas such that I can construct them myself (at least for guys with complete major-league records)?