Decided I should respond to John Murphy's question here, rather than in the discussion of the 1927 ballot results . . .

Are there any pitchers from Rube Foster's time that are comparable to him. I'm trying to find out where he belongs among the McGinnity's, Waddells, etc., for career and peak. Thanks!

Well, I can give you some opinions, anyway.

The other top black pitchers during Foster's career were Danny McClellan, Walter Ball, and Harry Buckner. It's pretty clear to me that Foster was the best of this group, though Buckner's career is so poorly documented (he began pitching for major teams in 1896) that it's imaginable that he was better than Foster. Foster's reputation is by far the strongest of this group, however.

In terms of value, I myself see Foster as pretty similar to Ed Walsh and Three Finger Brown, both of whom we elected quickly. Neither would have made it into my PHOM yet, if I had one, though they probably would before 1933. Foster is clearly not ranked similarly to Walsh and Brown by most of the electorate . . .

Foster's career shape is more like that of short-career, high-peak pitchers like Walsh, McGinnity, and Waddell than of a steadier producer like Brown. I see his career shape as being somewhere between Walsh's and Waddell's. Waddell had only about three tremendous seasons; Walsh had a seven-year run. Walsh had very little outside those seven years; Waddell had almost ten years of good pitching. Foster had a 6 1/2 season peak -- 1903 to the middle of 1909 when he broke his leg, with about 3 league average type years in 1902, 1910 & 1911, and three below average years 1912-1914.

I don't have time this morning to run through the specific conversions of Foster's available stats to MLEs and estimated win shares (I did those by hand and never typed them up) because I'm leaving this afternoon for a long weekend away, but if you'd like to see that data, I can work it up next week.

Thanks, Chris!

I was going to try to put together my case for Foster this week, but really I don't have much to add to what Chris says here, and what has been said in the discussion threads.

My take of the evidence is that his 1903-1908 are 6 star or superstar seasons, consecutively, which immediately makes him better than any pitcher we currently have on the ballot. I view him a lot like Jennings and Caruthers (who were 2 and 3 on my last ballot to Foster's 1): A ton of value in a short period of time. The reason he ranked (and ranks) #1 above those two is that I think his 1902, half of 1909, and 1910-11 comprise more value outside of the peak than the other two.

I know there is a line of thought that other pitchers were as good as Foster in that era of the NL; if there is evidence of that, I'd be glad to hear it. The line of thought is that they all were about the same, but Foster ahs more pub based on his later accomplishments. But Foster also has a number of contemporary admirers, and those cannot be explained so easily by his role as a builder.

Chris - I'd love to see your approximation on his MLE's and Win Shares; I'd similarly like to see him run through JeffM's similarity system afterward. Like you, I think a comp to Ed Walsh is quite appropriate. Let's see if your numbers say the same thing.

Chris - I'd love to see your approximation on his MLE's and Win Shares; I'd similarly like to see him run through JeffM's similarity system afterward. Like you, I think a comp to Ed Walsh is quite appropriate. Let's see if your numbers say the same thing.

I'll try to track down and type up my take on Foster's win shares over the weekend.

I should also note that, for those who find the i9s projections helpful, that they have recently (as far as I know) expanded their projections for pitchers to include ERA for many of them, including Foster, Mendez, and Donaldson. I just looked at their projections for Foster last night, and I haven't worked with them yet, so no conclusions to draw.

Having studied the i9s projections for Rube Foster and re-done my analysis of the available date, I'm inclined to think that i9s underestimates Foster's value somewhat.

Working with the i9s numbers, I derive for him a career ERA+ of 108.9, which works out to a support neutral record, given the innings they project (which seem reasonable), of 223-188 (wp .543). That's very good, but borderline HoM, projecting to about 275 career win shares in 3844 innings pitched.

In my view, they underrate Foster's dominance at his peak. For 1903-1907, they give him ERA+ totals of 123, 125, 99, 125, and 143. It looks to me like their numbers for Foster (and perhaps other pitchers) should be adjusted upwards by about 5%, while their numbers of batters should be adjusted downwards about 5%. In Foster’s case, I think the adjustment should be about 10%, but he’s the first pitcher I’ve studies, so he may not be representative, or I may be overestimating Foster’s ability. But there’s certainly no evidence that i9s is overestimating, so I would strongly argue for Foster being ranked at least as high as the i9s projections would indicate.

Based on numbers from Riley, Holway and Cottrell's biography, I give Foster ERA+ totals of 156, 133, 120, 130, and 112 for these peak seasons. (The analysis that gets to these numbers is long, but I can write it up if anyone is interested).

From 1908-1914, i9s sees Foster as pretty much an average MLE pitcher in terms of effectiveness and durability. Again, I see him as slightly better: he's average in most seasons, but he's a notch above average in 1908-1910.

So, using i9s (mostly) for projected innings and my own estimates of ERA+, here's my view of Foster's career as a pitcher

(no batting credit included)

season -- IP -- ERA+ -- SN wp -- WS (comments)
1902 -- 316 -- 105 -- .524 -- 19 (great stuff, but raw rookie)
1903 -- 378 -- 156 -- .709 -- 46 (totally dominating*)
1904 -- 350 -- 133 -- .638 -- 37 (great year)
1905 -- 325 -- 120 -- .590 -- 30 (1.66 ERA, Cottrell shows**)
1906 -- 326 -- 130 -- .610 -- 33 (close to 1904)
1907 -- 312 -- 112 -- .554 -- 23 (8-5 in Cuba)
1908 -- 298 -- 112 -- .554 -- 22 (about like 1907)
1909 -- 156 -- 104 -- .520 -- 10 (broken leg, rough return)
1910 -- 250 – 138 -- .656 – 28 (Holway shows 2.17 RA***)
1911 – 209 – 98 -- .490 – 11 (Rube’s first below-average year)
1912 – 271 – 100 -- .500 – 16 (average season; Holway shows 4.58 RA)
1913 – 258 – 100 -- .500 – 15 (average season)
1914 – 246 – 105 -- .524 – 16 (a bit above avg. Holway shows 3.38 RA)
1915 – 121 – 93 -- .463 – 5 (spot starts)
1916 – 50 – 84 -- .413 – 2 (token appearances)
1917 – 31—84 -- .413 – 1 (token appearances)
tot – 3897 – 117 -- .578 – 314 career win shares
est. support neutral record 241-176


*Riley reports 54-1; Holway shows Foster 4-0 in championship series against Philadelphia Giants with an RA of 1.50, which suggests an ERA of around 1.00.
**This ERA, with other data from this exceptionally well-recorded season, is instrumental in setting various estimates for the early part of Foster's career.
***Foster appears to have risen to challenges throughout his career; it looks like he raised his game a notch the first year he had full control of a team)

Final Comment

Given the extant data, I’m pretty confident this projection captures Foster’s ability as a pitcher. The matter I’m most uncertain about is the number of innings that Foster threw 1910-1914. i9s sees him as a pitching the load of a 3rd starter; Foster certainly was pitching less during these years, but how much less is a bit unclear.

Wow! Great work Chris! Foster's shooting up my list this week.

To try to put his year-by-year in more generic numbers, here's his year-by-year W/L (figured by dividing the season's IP by career IP, multiplying by career decisions for the number of decisions he should've had that year & then multiplying by SNWL for the season W/L record (keep in mind this is a SNWL, not a normal W/L record where support is almost never neutral):

1902..17-16
1903..28-12
1904..24-13
1905..21-14
1906..21-14
1907..18-15
1908..18-14
1909..9-8
1910..18-9
1911..14-15
1912..15-14
1913..14-14
1914..14-12
1915..6-7
1916..2-3
1917..1-2
Total: 240-182

Also, Rube Foster could freakin' hit. When I saw his hitting stats I was reminded a little bit of Bob Caruthers, so SNWL underestimate him because he was so good at supporting himself.

Other comparisons: not nearly the peak of, say, Ed Walsh, but much more post-peak value.

Wow, that's amazing work, Chris!

My initial impression: Slightly less valuable in 1907-1908 than I had previously thought, but significantly better in 1910 than I'd thought. Also, as Chris J. said, his offense adds a little to his game around the edges.

The Chrises,

Any way you could run the numbers for John Donaldson?

Hmmn - he's even better than I thought he was - he'll be moving up for sure this week.

Everyone should be looking at this . . .

Two Notes offering revisions to Foster's estimated win shares and support-neutral winning percentage from post 5 above:

1) My work on Jose Mendez this week brought to my attention the fact that my system for estimating win shares from support-neutral winning percentage appears to work well for values up to the upper 120s, but above that it begins to skew values upwards: the higher the ERA+, the stronger the distortion. I compensated for that by applying discounts to Mendez's win shares in his top seasons. Going back and looking at my estimates for Foster, I see that the problem certainly arises in his top seasons as well.

I thus suggest the following reductions, which I will use in placing Foster myself:

1903 -- Drop from 46 to 40
1904 -- Drop from 37 to 34
1906 -- Drop from 33 to 31
1910 -- Drop from 28 to 26

That drops Foster's estimated career win share total for pitching from 314 to 301 and drops the height of his peak somewhat. It doesn't change his support-neutral w/l record as Chris J. has calculated it, but it affects its value.

I hope anyone relying on my estimates will take this revision into account, since I believe it to be a more reliable estimate.

2) My work with early pitchers has convinced me that the win shares system overrates the value of an average pitcher in all eras prior to the development of the modern home-run offense during the 1920s. To make Foster (and Mendez) comparable to their peers as represented by win shares, I calculated their win-share values according to win shares' fixed rates. I thought publicly available win share estimates ought to be as close to what the actual win share system would produce as possible (part of the reason for my first suggested revision above).

However, in my view, these rates overestimate the value of both pitchers somewhat, and I want at least to offer the values that I will be using to rank them.

Here's the formula. Win shares says that an average pitcher should earn .058 ws/ip . I recalibrate this as follows. For 1901-1910, an average pitcher earns .0477 ws/ip; for 1911-1920, an average pitcher earns .052 ws/ip; for 1921-1930, an average pitcher earns .0546 ws/ip.

Using this formula, Foster's career pitching win-share total drops by 35 from 301 to 266. Those interested in season-by-season adjustments can do the math easily enough. . .

Since I apply this discount to all pitchers, it doesn't affect Foster's standing much in my rankings in relation to other pitchers, but it does explain why I have ranked pitchers fairly conservatively over the years.

Going by earlier discussions, I expect that many voters will disagree with this adjustment, but I thought, especially since there have been some concerns expressed about rushing Foster in, that I should pitch a more conservative view of pitching value 1900-1920.

Having made these downward adjustments, I would then adjust Foster upward by adding batting win shares (he was a very good hitter -- I see him as comparable to Walter Johnson) and by adjusting for season-length.

Even with these downward modifications, Foster is still a strong candidate. I see him as deserving of election. But I don't believe our failure to elect him yet has been an oversight of monumental proportions. I'd have put him in before Brown and Walsh, but I wouldn't have elected Brown yet.

Ok, I just don't get the Rube Foster wave of support.

He's supposed to be a surefire HoMer based on a five-year run of 1556-133-120-130-112 ERA+.

Well, there are dozens of guys better.

For example, Hippo Vaughn 1916-1920, 132-144-161-161-126, 6th-5th-1st-1st-5th in IP.

Foster is in the mix with Griffith, Willis, Waddell, Joss, Cooper, Vaughn, etc. I suppose, but is towards the bottom of that group (without credit for his hitting). Giving him significant credit for his hitting, I've got him somewhere around 20th.

Incidentally, Vaughn and Cooper are wildly underrated by our electorate.

For me, Foster is positioned highly in that group because of career value, not because of peak. He has more career than anybody in that group except Willis, whom he does beat on peak, I think.

Since Cooper hasn't been voted on yet, I don't know if he will be underrated or not, but he hasn't been much discussed, so it looks like he may not get much support. He's going to make my ballot, fwiw.

Andrew, do you see Vaughn as ballot-worthy? He did have a great run 1916-1920, but his career was rather short. I've realized that I didn't give him serious consideration, but haven't had time yet to study his record thoroughly.

For me, Foster is positioned highly in that group because of career value, not because of peak.

That's how I interpreted your figures, Chris. Foster wasn't a peak monster by any sretch of the imagination, but coupling his peak with his career numbers makes him ballot worthy, IMO (though not HoM bound at the present time).

Rube Foster's Real Stats

1905 MMP Prelim:

1) Honus Wagner
2) Cy Seymour
3) Christy Mathewson
4) Ed Reulbach
5) Mike Donlin
6) Rube Waddell
7) Roy Thomas
8) Rube Foster
9) Eddie Plank
10) John Titus

Heh. Wrong thread. :-)

It's all good, John. I was going to go digging for this in the archives for my MMP vote and you saved me the time. Thanks.

Hey, guys and gals,

Please find my latest MLE for Rube Foster here. The fully articulated MLE method is linked in there.