I’ve read some discussions about greatest single season pitching performances, but I never really contributed because I never looked into the matter. Thanks to last week’s blackout, I did some fooling around with Support Neutral Stats. Michael Wolverton developed them and they are availble on BPro.  BPro has Support Neutral stats from 1960 on (I think they’ll be able to go deeper into baseball history as retrosheet adds more info. I took the top 38 Adjusted ERA+ seasons since 1960 (those were the ones that made bbref’s top 100 leader board.) Looking at ERA+ and support neutral stats, Bob Gibson had @ the 3rd-6th best season in that timeframe back in ‘68. The best? It depends how you skin that cat. It may be Koufax in ‘66, Pedro in Y2K, or Gooden in ‘85.

I looked at ERA+, SNVA, and the likelihood that a .500, .425. or .350 pitcher would have one of those top seasons.

Four of the top 5 seasons were done in less than 30 starts.

If I used Excel correctly, those are the top 10 seasons in expected team wins and losses using support neutral stats. I ranked them three times: once for the probability that a .500 pitcher would have a season like that and once for a .425 or .350 pitcher. If you use average as a baseline, than Pedro still had the best season in ‘00. If you use a replacement level value (Wolverton used .425. I also included .350, as I’ve seen elsewhere), then Koufax pulls ahead.

SNVA. If those past three rankings measure ability, this may be more akin to value. Using a .500 team as a baseline, they’d be 89.5-72.5 if they added Koufax, 89.4-72.6 with Gooden, &c. Basically a wash between the top 2.  Actually, I should have probably also calculated Support Neutral Wins above Replcement and combined that with SNVA.  IIRC, you get augmented returns for each extra win above average; if you are looking at this from a Pennant Expectancy perspective (and that’s the ultimate goal.)

These just scratch the surface, but I had a night to kill and this is what I came up with. Is this the final word on the subject? Hardly. There are alot of things unaccounted for like defense, catchers, coaches, and managers. Not the mention the fact that it considers every game relatively similar and doesn’t account for a possible start where the guy has an 11-0 lead and takes one for the team by going deep into the game and giving up 6 runs (for example).

If this warrants further study, I may try and get to it. If not, I just spent a night brushing up on my Excel skills.

To clarify the previous post, here’s an example of the report that I used.  The items in the header row link to BPro’s glossary, if you’re unfamiliar with what they mean.  That’s the 1975 report.  I would like to hear comments on the validity of ranking players by the probability that an average or replacement player might be able to replicate the season. I know that Bill James used that methodology in 1983 and 1984, if not before, but he dropped it for another ranking system and I never found out why.