Posted by doubletruck on 1/13/2015 5:32:00 AM (view original):
I saw Koufax, Johnson, Spahn and Carlton pitch, several times each. I voted for Koufax, but if you insist on a long career, then it's Johnson (with Spahn and Carlton next in line).

"Greatest" doesn't mean longest to me. Koufax and Johnson both started slow, but for one five-year span Koufax was consistently untouchable.

The talent was more concentrated in the 1960s and pitchers didn't count on turning the game over to the bullpen for the late innings. Koufax had 137 complete games in a third fewer games pitched than Johnson, 137-100.

Of course, if you count birds killed in flight, then it's Johnson all by himself.

Johnson's best years were better than Koufax's best years.

What stats did you cherry pick to come up with that conclusion?

Posted by bad_luck on 12/31/2014 2:44:00 PM (view original):
Career        IP        SO       ERA      ERA+     Whip    WAR
Koufax       2324   2396     2.76      131       1.106     53
Johnson    4135   4875     3.29      135       1.171    104


Best 3 seasons    IP      SO      ERA     ERA+    Whip   WAR
Koufax                  881    922     1.85     176       0.92      26
Johnson               758    1053   2.48     188       1.052    29

And if you stretch it out to best 5 or 7 years overall, Johnson runs away with it. To argue that Koufax was better than Johnson, you have to ignore the fact that the run scoring environment in the late 1950's and early 1960's was significantly different than it was in the 1990's and 2000's. And then you have to ignore the fact that Johnson doubled up on Koufax's IP.

Pretty sure it was this...

To be fair, if you want to look at it from the perspective of ERA+, which isn't entirely unreasonable, Johnson is the clear winner.  Koufax had 8 seasons in which he qualified for the ERA title.  Compared to Johnson's 8 best seasons, you can pick any sample size of best years you want and Johnson comes out ahead:

Johnson			Koufax
Season	Age	ERA+	Season	Age	ERA+
1997	33	197	1966	30	190
2002	38	195	1964	28	186
1995	31	193	1965	29	160
2001	37	188	1963	27	159
1999	35	184	1962	26	143
2000	36	181	1961	25	122
2004	40	176	1960	24	101
1994	30	152	1958	22	93

That being said, as I've said in the past, even + stats can't necessarily do an adequate job of normalizing eras.  Johnson played in one of the most offensive eras in league history.  That means the numbers are bigger, and the variance is bigger, so it's easier to have slightly more impressive + stats, especially as a pitcher.  Koufax played in the least offensive period of the live-ball era, making it very difficult to achieve the same ERA+ numbers.  I mean heck, his best ERA seasons were 1.73 and 1.74.  It's hard to be much better than that; anybody human is going to make mistakes, and in live-ball baseball a few mistakes add up to a high-1s ERA at best.  Since 1920 only 8 pitchers have combined to put up 9 seasons with better ERAs than what Koufax had in '64 and '66.  Probably Hubbell doing it in 1933 is more impressive, or Maddux in '94 and '95, or Kershaw being only slightly worse the past few seasons.  But still, those numbers are so far to the high end, I think they might merit a little more appreciation than the + stats would imply, because of the run scoring environment in which they occurred.

Posted by doubletruck on 1/15/2015 3:55:00 PM (view original):
What stats did you cherry pick to come up with that conclusion?

It's pretty easy. Look at Koufax's best year (1966?) and compare that to Johnson's (1997 or 2002 are about the same). Johnson was slightly better. But it's close. Then look at the second best for each. I'd probably pick 1964 for Koufax but 1965 is also a good choice.  Then pick the second best year for Johnson. Again, Johnson was slightly better.

You can keep going, but Koufax runs out of full seasons (only 9 where he threw more than 150 IP) before you get to a full Johnson season with an ERA+ less than 135.

Maybe ERA+ doesn't tell the full story (it doesn't). But Johnson beats him in WAR, too. Using unadjusted ERA and WHIP favor Koufax but both ignore the fact that the run scoring environment in the 1960's was significantly different than it was in the 1990's and 2000's.

Posted by dahsdebater on 1/15/2015 4:20:00 PM (view original):
To be fair, if you want to look at it from the perspective of ERA+, which isn't entirely unreasonable, Johnson is the clear winner.  Koufax had 8 seasons in which he qualified for the ERA title.  Compared to Johnson's 8 best seasons, you can pick any sample size of best years you want and Johnson comes out ahead:

Johnson			Koufax
Season	Age	ERA+	Season	Age	ERA+
1997	33	197	1966	30	190
2002	38	195	1964	28	186
1995	31	193	1965	29	160
2001	37	188	1963	27	159
1999	35	184	1962	26	143
2000	36	181	1961	25	122
2004	40	176	1960	24	101
1994	30	152	1958	22	93

That being said, as I've said in the past, even + stats can't necessarily do an adequate job of normalizing eras.  Johnson played in one of the most offensive eras in league history.  That means the numbers are bigger, and the variance is bigger, so it's easier to have slightly more impressive + stats, especially as a pitcher.  Koufax played in the least offensive period of the live-ball era, making it very difficult to achieve the same ERA+ numbers.  I mean heck, his best ERA seasons were 1.73 and 1.74.  It's hard to be much better than that; anybody human is going to make mistakes, and in live-ball baseball a few mistakes add up to a high-1s ERA at best.  Since 1920 only 8 pitchers have combined to put up 9 seasons with better ERAs than what Koufax had in '64 and '66.  Probably Hubbell doing it in 1933 is more impressive, or Maddux in '94 and '95, or Kershaw being only slightly worse the past few seasons.  But still, those numbers are so far to the high end, I think they might merit a little more appreciation than the + stats would imply, because of the run scoring environment in which they occurred.

Koufax was also pitching on a higher mound.

I feel like that information is already captured in the comparison of run-scoring environments...

I feel like you're saying (in your longer post), "Johnson pitched at a time when there were a lot of really ****** pitchers, making his seasons look more impressive, by comparison, than they were."

Am I reading that wrong?

Yes, you are reading that wrong.

That's your explanation for the offensive explosion of the 'mid-90's through mid-'00s, not mine.  Although it's an additional point of some significance if you believe it to be true - an ERA+ in a league which you yourself just described as unadjusted to the number of teams and stretched far too thin on pitching should certainly be downgraded a bit relative to an ERA+ from a league of largely established pitchers.

That's why I asked. I didn't want to assume.

If you don't think that pitching was diluted, what makes you think that ERA+ doesn't accurately reflect the dominance of each pitcher?

I don't think ERA scales linearly across the scale; ERA+ assumes that it does.  I think ERA+ works just fine (ignoring fluctuations in the overall talent pool) for guys in the general neighborhood of 100.  As you go to the extremes I don't think the same proportionalities relative to the average hold.  For guys who are very, very good - historically good - I think you need some combination of ERA and ERA+ to tell the whole story.  If you look at the best ERA+ seasons for Koufax then translate them to Randy Johnson's prime, they result in an ERA of ~2.5, or even a little bit higher, even in the National League.  However, as I referenced above, I don't think that's a realistic translation.  At the point where your ERA is in the 1.7s, you aren't making very many mistakes.  I just don't think you'd see that big an ERA jump with an equivalent performance.  I'd guess maybe more to 2.1 or 2.2; a little more punishment on the mistakes, but not too much damage.  And if you translated Johnson back into the 1960s, I don't think his ERA drops lower than maybe 2.0 or 2.1 in his best years.  Obviously this is all conjecture, and you can't really test it meaningfully.  But Pedro and Maddux proved you could utterly dominate the league in the '90s and early '00s, and Johnson didn't quite come up to that level.  I'd call Koufax in his prime somewhere between Maddux/Pedro and Johnson, but still better than Johnson.

Genuinely not being a dick:

What do you mean by, "I don't think ERA scales linearly across the scale."

This could be an interesting hypothesis.

The ML average ERA increased nearly 50% from the mid-60s to the mid-00s.  Part of that is attributable to the advent of the DH; I'm sure I could look up the NL stats (you can get year by year league ERAs at Fangraphs easily), but let's assume it's 40%.  Or even 30%.  It doesn't matter.  I don't think the ERA of a pitcher with an ERA around 2 increases by that same margin.  I do believe that a guy with an ERA of 3 translates, on average, to roughly 4.3, or whatever it is.  But I don't think 2 goes to 2.8.  I think it goes to more like 2.4 or 2.5.  If you're pitching that well, guys are mostly not going to hit you, regardless of the league environment in which you're pitching.

Pitcher A -- ERA of 2.50, ERA+ of 200..............Pitcher B -- ERA of 2.00, ERA+ of 200     bad_luck, do you consider these pitchers to be equivalent (assuming the same IP, etc)?

At 6-10, Johnson was pitching downhill as much or more than Koufax did on the higher mound ...