Koufax’s peak

Having just spent an entire post “putting down” Sandy Koufax, I’d thought I’d follow up by examining the question of whether Sandy Koufax’s peak was the best of all time. Instead of referring to WAR or WSAB, I decided to use “Situation-Adjusted WPA” or “Game-state adjusted wOBA” or some mouthful like that. It’s just WPA/LI.

I’ve learned to mistrust WPA for starting pitchers, for many of the same reasons that I mistrust a pitcher’s win-loss record. A team’s offense has too much impact on it. But WPA/LI quantifies how well the pitcher pitches to the situation he’s given, and one of the objections I heard about my Koufax post is that he did a good job of pitching to the situation. So WPA/LI should address that objection.

The problem with using WPA/LI is that we only have stats from 1954. So this post will address the question of the best pitching peaks since 1954.

The best single season of WPA/LI since 1954 was Pedro Martinez’s 2000, when he was 18-6 with a 1.75 ERA. His WPA/LI of 8.5 was more than a “win” higher than Bob Gibson’s 7.3 in 1968. Rounding out the top five at 7.2 are Roger Clemens (in 1997 with Toronto), Greg Maddux (1995) and Koufax (1963). Pedro was beyond amazing. To my surprise, Dwight Gooden’s 1985 was ninth (6.5).

The best two seasons (not necessarily consecutive) were also Pedro’s: 15.2 in 1999 and 2000. The next best two-season peak was Koufax’s in 1963 and 1966: 14.4 (7.2 in each season), followed by Maddux (13.5), Gibson (13.2) and Clemens (12.7).

The best three-season peak also belongs to Pedro, though Koufax is closing the gap. Throwing in Pedro’s 1997 brings his three-year total to 21.5, while Koufax’s total at three seasons is 21.2, by virtue of his outstanding 1965. The other best three-year peaks are still Maddux (19.8), Gibson (18.1) and Clemens (18.0).

Koufax takes over at four years, at 26.8. Pedro is second at 26.5. Maddux is third at 26.0, Clemens is fourth at 23.3 and Randy Johnson moves up to fifth place at 23.2. Gibson is sixth.

The all-time five-year peak season leader is Greg Maddux. Maddux has 31.4 “situational wins” (I’m not really sure what to call them) over his top five seasons, which also happen to be consecutive (1994-1998). Pedro is just barely second with 31.3 (mostly 1997-2003, but skipping a couple of years) and Koufax is third with 30.5. Clemens is fourth, Johnson fifth and Gibson sixth. Rounding out the top ten are Kevin Brown, Tom Seaver, Juan Marichal and Jim Palmer.

Once you go beyond five years, I don’t think you’re really talking about a peak anymore, but Greg Maddux just sort of takes over.

So was Koufax’s peak the best of all time (actually, since 1954), even adjusted for his low run-scoring environment and situational pitching? He’s certainly right up there with Pedro and Maddux. I don’t have enough faith in the metric to declare one or the other “the winner,” but if you want to call his peak the best in the second half of major league history, I won’t object.


Print This Post
Dave Studeman was called a "national treasure" by Rob Neyer. Seriously. Follow his sporadic tweets @dastudes.
Sort by:   newest | oldest | most voted
Lou
Guest
Lou

wish we had the data to see Grove and Walter in this

Silver King
Guest
Silver King

Cool stuff!  Thanks

SharksRog
Guest
SharksRog
Sandy was a much different pitcher in Dodger Stadium than he was in the other stadia in which he pitched. I would be intrigued to see how Sandy’s WPA/LI numbers break down between Dodger Stadium and on the road.  In actuality, I wouldn’t expect to see much difference.  Sandy pitched a lot better in Dodger Stadium, but so did the opponents’ pitchers. I think both Sandy’s WAR and his VORP would show a very large difference between Dodger Stadium and on the road though. Because of the hitters’ park nature of Ebbetts Field and the LA Coliseum and the pitcher’s… Read more »
studes
Guest
studes

SharksRog, read my post from yesterday.

WPA/LI takes the run environment into account.  Whether it does so perfectly is another matter.

gary
Guest
gary

These studies generally don’t adequately address Kiufax’s huge IP advantage. He had 350 more IP than Pedro. If you cherry picked his best 1022 IP and compared it to Pedro’s 1022 IP – now that would be interesting.

Dave Studeman
Guest
Dave Studeman

Interesting point, Gary. OTOH, achieving a higher WPA/LI in less innings is pretty impressive.

Bob Sanchez
Guest
Bob Sanchez

Well, if you cherry-picked any pitcher’s best IP, you could just stack up perfect innings!  I’m not sure what that would prove.  It would be interesting to see which pitcher has the most!

Cyril Morong
Guest
Cyril Morong

I have done some research on a related note. It is not as sophisticated as what Dave has done since it is not clutch-based. But I did not find that Koufax had the best peak.

Bert Blyleven: As Dominating as Sandy Koufax

http://www.beyondtheboxscore.com/story/2007/1/31/8555/76382

How Good Was Sandy Koufax Outside of Dodger Stadium? (I compared him to Gibson, Marichal and Bunning)

http://www.beyondtheboxscore.com/story/2006/5/5/12349/01432

The Best Five-Year Pitching Performances Since 1920 Based on Fielding Independent ERA

http://www.beyondtheboxscore.com/story/2006/7/17/9537/72813

The Best Five-Year Pitching Performances

http://www.beyondtheboxscore.com/story/2006/5/1/94552/53377

Tim
Guest
Tim
I know that the traditional view is that a pitcher on a contending team has an advantage over one on a loser but I would like to submit something different. From 1910 through 1927, Walter Johnson rather famously pitched for losers. Over that period his Senators averaged finishing out of 1st by 18.66 games. Some of their worst years they were out of contention as they broke from Spring Training. Now according to the traditional view a pitcher would have a difficult time posting a great record under those conditions but if the pitcher is a great talent as Johnson… Read more »
gary
Guest
gary

My point was that Koufax threw an additional 350 more IP than Pedro during his 5 year peak. Pedro’s superior rate stats have to be somewhat offest by that. I mean we are talking 70 IP a year more for Koufax. That’s Pedro plus a 70 inning a year reliever. Each player had some advantages to the era they pitched in, but I would give Koufax the slight edge in peak based on the IP disparity.

Dave Studeman
Guest
Dave Studeman

It’s a good point, gary.  I was thinking about how to correct for that—one thing to do would be to add innings to Pedro’s total, assuming they regress to some sort of a WPA/LI/IP level based on the two years before and after the year in question.  Something like that.

Actually, the best thing to do would be to set WPA/LI against a replacement level instead of against average.  That would take some work…

gary
Guest
gary

It looks like RSSA will always favor pitchers from high scoring eras because those from low scoring eras will not be able to get enough separation from their peers.

During Pedro’s 1997-2001 peak, 13 pitchers logged more IP than he did. He pitched fewer IP than Brad Radke, Darryl Kile, Aaron Sele and Jon Lieber – just to name a few. Also, 38 pitchers started more games than Pedro from 1997-2001. Koufax ranked 4th and 8th respectively in IP and GS during his peak.

SharksRog
Guest
SharksRog
No question Sandy Koufax pitched a lot more innings per season in his peak than did Pedro Martinez.  The game is different now. Sandy’s innings during those five years pale compared to those of Walter Johnson.  In the first five years of Walter’s peak, he averaged 80 more innings per season than did Sandy. In his entire 10-year peak, Walter averaged almost 70 innings per season more than Sandy. It was a different game. Meanwhile, the run environment during Sandy’s peak was somewhere around halfway between the run average during Walter’s peak and Pedro’s peak. Many will argue on Koufax’s… Read more »
Dave Studeman
Guest
Dave Studeman

It looks like RSSA will always favor pitchers from high scoring eras because those from low scoring eras will not be able to get enough separation from their peers.

@gary, you should read the comments in Saturday’s post, as well as the link to the article about ERA+.

gary
Guest
gary

SharksRog,

But Pedro didn’t even dominate within his era in terms of IP and GS. By comparison, Grove led the majors in IP from 1929-33 and was 5th in GS.

Yeah, we all get it – Koufax pitched in a pitcher friendly environment and he must be punished for that and Pedro pitched in a hitter friendly era so he must be endlessly praised. Seems rather simplistic to me, but whatever.

gary
Guest
gary
Dave, I hate to sound like an anti-stat guy, but any methodology that ranks Roy Oswalt above Walter Johnson doesn not pass the smell test. ERA+ definitely favors high scoring era pitchers. And even it it does not, we are ignoring the fact that preventing a run in 1966 is more valuable to a team than preventing a run in 1997. The bottom line for me is that I don’t think any one formula is going to tell us whether Pedro 1997-2001 was really better than Koufax 1962-66. Certain trendy stats favor Pedro, and perhaps he was better, but to… Read more »
SharksRog
Guest
SharksRog
Gary— Ignoring everyone but Sandy and Pedro now, let’s look at some of the various issues: .  You make a very good point that Pedro wasn’t among the leaders in innings pitch, while Sandy led the NL in innings pitched in two of his five peak years.  Advantage Sandy. .  Pedro pitched in a much tougher hitters’ era.  Advantage Pedro. .  Pedro pitched in a much tougher hitters’ park.  Advantage Pedro. .  Sandy had the lower ERA of the two.  Advantage Sandy. .  Pedro’s peak lasted seven years, Sandy’s five (or actually 5 1/2 if one looks at the second… Read more »
SharksRog
Guest
SharksRog
By the way, I think I would take Walter Johnson’s much longer peak over either of them.  Greg Maddux had a long peak, as well. But over a seven-year period was there ever a pitcher better than Pedro WHEN PEDRO pitched (giving that he was injured for half a season during that time and didn’t dominate in innings as Johnson and Koufax did)? Given the era and the home parks, I’m not sure there has been. I don’t mean to be bashing Sandy here, but there is considerable evidence that in the innings each pitched (which takes away perhaps Sandy’s… Read more »
Dave Studeman
Guest
Dave Studeman
ERA+ definitely favors high scoring era pitchers.  Did you read the article? A more helpful comment would highlight why you think the article is wrong. Oswalt and Johnson are virtually even in both ERA+ and mERA.  If you don’t think he belongs up there, you shouldn’t like ERA+ either. Don’t forget that these are rate stats and don’t account for length of career. Also, don’t forget that we’re always talking about relative value here.  We’re not trying to guess how so-and-so would have done in a different environment. We’re trying to compare value in different environments. You’re right that not… Read more »
SharksRog
Guest
SharksRog
It’s darn tough for Sandy to compete with longevity lefties such as Warren Spahn, Steve Carlton, Carl Hubbell and Lefty Grove for career, but I do believe that with the possible exception of Grove, he was better than any of them for peak. There are so many things to consider with Sandy: .  What if he hadn’t been a bonus baby and had been able to develop in the minors? .  What if Norm Sherry (IIRC) had been around earlier to get Sandy to throw a little less hard in exchange for far better control? .  What if Sandy hadn’t… Read more »
gary
Guest
gary

Koufax was not exactly chopped liver on the road. His road ERA from 19622-66 was tops in the majors.

A more fundamental question – would it ever be possible for a pitcher in a low scoring environment to ever lay claim to a best peak argument? The deck seems stacked against them and in favor of guys who pitch in high scoring environments.

SharksRog
Guest
SharksRog
Sandy was excellent on the road during his peak.  He was unbelievable at home. I see no reason why a pitcher in a low-scoring environment could’t have the best peak. Between Pedro (high-scoring era), Sandy (medium-scoring era) and Walter (low-scoring era), I would give both Walter and Pedro the edge over Sandy for five-year peak (although all three—and a few others—are very close IMO). For seven years, I would probably go with Pedro.  For a full decade, it seems to easily be Walter (at least among this trio). So among the trio, both the guy in the low-scoring era and… Read more »
wpDiscuz