Thinking about this again, with Felix Hernandez winning the AL Cy Young today.
I've recently become obsessed with the idea that the greatest non-strike-shortened, single-season achievement in baseball is for a pitcher to win over 20 games and lose two or fewer. It's never been done. Never.
Two pitchers came very close in 2008: Cliff Lee, then of the Cleveland Indians, and Daisuke Matsuzaka of the Boston Red Sox. Lee finished 22-3, and Matsuzaka 18-3. Lee didn't get his third loss until a late-September matchup with the Red Sox, and Dice-K didn't have a realistic shot at 20 wins, as other nuances of his game left him with eight no-decisions for the season (he didn't get loss #3 until September 28, just for the record).
Let's examine Cliff Lee's season for a second. His ERA is telling: in his 22 wins, it was 2.06. In his three losses, it was 6.56. OK, everybody has a bad day now and then. But in his no-decisions, it was 2.57. There was a game when he pitched nine innings of shutout ball, only to get yanked at the end of nine, tied 0-0, with the Indians losing it in the 10th. That's Harvey Haddix territory.
Now let's compare him to Matsuzaka. Dice-K's ERA was as follows: in his 18 wins, it was 1.60. His three losses? An ERA of 14.40. And in his no-decisions... 3.57.
So what does any of this mean? To me, it means that Dice-K got shellacked in his losses, but probably should have won one or two more games, as he was decent in his no-decisions and so was actually a lot better than his already great record. It also means that Cliff Lee was excellent in 28 of his 31 starts, which means he could have hypothetically gone 28-3, eclipsing Ron Guidry's insane 1978 season.
So let's take a look at Ron Guidry's completely unbelievable 1978. Guidry went 25-3, receiving loss #3 smack in the middle of a five-start stint in September where he either threw a complete game or a complete-game shutout. Win 25 came in the one-game play-off with the Red Sox, and I can tell you that as a Red Sox fan, it hurts more to believe that Bucky F. Dent won the game for the Yankees rather than Ron Guidry In The Midst of His Cy-Young-Award-Winning Season.
Guidry's season ERA was a paltry 1.74: his ERA in wins was 1.11; 4.99 in losses; and 3.47 in no-decisions. He made 35 starts, which is four more than Lee, and six more than Dice-K. Two things are interesting to me here: 1) he wasn't lights-out in his no-decisions (unlike Lee); and 2) the probability of a Yankee loss during a Guidry no-decision was probably minuscule, as Hall of Famer Goose Gossage was in the bullpen in 1978, in the midst of his own lights-out season.
Let's address point 2) first. Gossage participated in six of Guidry's seven no-decisions. Goose won three, lost two, and had at least one blown save. I'm not very good with figuring out percentages, but Gossage got three of his ten wins during Guidry no-decisions, which should mean something in terms of scaring the opposition.
Now, back to Guidry's ERA. Why am I so hung up on a pitcher's ERA? Because it's the average run total the pitcher staked his opponent. The 1978 Yankees gave Guidry 4.85 runs in support in his starts. So it makes sense, then, that his high ERA matched his losses: 4.99 ERA to 4.85 run-support. Also that he gave up less in his wins: 1.11 ERA to 4.85 run-support. But his ERA in no-decisions: 3.47. The Yankees were still averaging more than a run more than this (4.85). So based purely on averages, Guidry should have won a few more games.
This same treatment yields Cliff Lee a few more wins as well. The 2008 Cleveland Indians averaged 5.58 runs in support of his starts, while his no-decision ERA was 2.57.
The final thing I can say about Cliff Lee's 2008 campaign is that he accomplished it for a completely mediocre team: the 2008 Indians went 81-81. He didn't have Guidry's luxury of pitching for a team on its way to the World Series, nor was he surrounded with a great rotation like Daisuke Matsuzaka was on the Red Sox (a lot of people thought Dice-K wasn't even the best pitcher on his team that season, that Jon Lester, Josh Beckett, or Jonathan Papelbon was, take your pick).
But this idea of only two losses ... it's the loss-column total that's so hard to master and leaves "only 3 losses" somehow a consolation prize to immortality.
I'm not including the magical 1995 season when Randy Johnson of the Seattle Mariners finished 18-2 and Greg Maddux of the Atlanta Braves finished 19-2 in this simply because it wasn't a full, 162-game season (it was 144 games).