Thursday, October 06, 2005

Pitching in a Pinch

Here, now, the Cy Youngs:

NL CY YOUNG: Roger Clemens

Boy, every award selection I make comes with a very complicated explanation, doesn't it? So I'll make this one very short: ERA is a much more reliable indicator of a pitcher's quality than Wins and Losses.
In brief: A Win is awarded to the pitcher who is in the game when his team takes the lead for good. A loss goes to the pitcher who allows that deciding run. In every major league baseball game, one pitcher will get a win and one will get a loss. It does not matter how well they pitch or how many pitchers pitch well; one gets a loss, one gets a win. People think that a guy with a 20-10 record pitched 20 good games and 10 bad ones. I challenge you to go to and count up good games and bad ones. The odds are astronomical that it will be 20-10; it could be 10-20 or worse or better. Pitchers get awarded a win or a loss all the time when they don't pitch that well or that poorly. In a game where two pitchers pitch brilliant, one of them will get the loss. And don't even get me started on bullpens; they screw up our ideas of good wins and bad losses completely. So Wins and Losses don't indicate how good a pitcher is, and if you think they do, you're a moron.
I know what you think: a pitcher who wins games with a bad ERA is still doing something right; he's a "proven winner." Well, can I show you how many "proven winners" went to bad teams and became "proven frauds"? A won-loss record says much more about a team's quality than the pitcher, especially in the age where starting pitchers pitch so few innings.
It takes (at least) 9 men to win or lose a baseball game. Giving all the credit or blame to one pitcher is just plain stupid. If you want to know how many good games a pitcher pitched, look at Quality Starts. ERA is imperfect; you have to adjust for ballparks and the league context, and even defense. A defense can have a strong impact on a pitcher's ERA, even without considering unearned runs. But it's the best (basic) pitching statistic we have.
Okay, so that wasn't so short, but now that I've established that, I'll tell you why I think Roger Clemens was the best pitcher in the NL. Here's a comparison of the three leading candidates:

Chris Carpenter (21-5, 2.83, 241.2 IP, 18 HR, 51 BB/213 K)
Dontrelle Willis (22-10, 2.63, 236.1 IP, 11 HR, 55 BB, 170 K)
Andy Pettitte (17-9, 2.39, 222.1 IP, 17 HR, 41 BB, 171 K)
Roger Clemens (13-8, 1.87 ERA, 211.1 IP, 11 HR, 62 BB, 185 K)

Clemens has an absolute gulf of an advantage in ERA. It looked a lot closer until Carpenter self-destructed down the stretch. It's virtually impossible that, ballparks aside, a pitcher with an ERA 1 full run higher is actually better. And Clemens pitches in a much tougher ballpark than either Willis or Carpenter!
The argument for Carpenter (barring the stupid one for wins) is that he pitched more innings with a better BB:K ratio. This is actually a pretty good argument, and if Carpenter's ERA were about 2.20 or lower, I might accept it. But Carpenter's small edge in IP (30.1 more than Clemens) and BB:K ratio (11 fewer walks, 28 more K) just cannot account for the gulf in ERA. A pitcher's job is to prevent runs from scoring, and Roger Clemens did that better than anyone else in the league by a substantial margin.
What about Dontrelle Willis? He's better than Carpenter, but still doesn't have enough to match Clemens. He has a better ERA (somewhat accounted for by his pitcher-friendly ballpark) in a similar number of innings, but a sizeable gap in strikeouts (43 less). While I might accept Carpenter over Willis as an argument, Willis is the better hitter (which may or may not matter in the voting) and Carpenter's stumble down the stretch can't be ignored. I also look at the 7 fewer HR allowed. I give Willis the edge over Carpenter, just barely.
But why did I put Andy Pettitte's name up there? He's not considered a serious candidate for the Cy Young? No, but he's had a better year than Carpenter and Willis in ERA, substantially so when you consider his home ballpark. He's allowed fewer HR than Carpenter (again despite the ballpark), and has a better BB:K ratio than Willis. Yeah, he's my #2.
Any other candidates? I actually will take Roy Oswalt over Carpenter (that September stumble sent him reeling), considering a similar ERA (2.94 for Oswalt) and similar peripherals despite Oswalt's ballpark. Any relievers to consider? Chad Cordero looked like a possibility (although his home ballpark made him look a lot better than he was), but both he and Todd Jones stumbled down the stretch. I think Billy Wagner was the best reliever in the league, but he wasn't Eric Gagne-good, i.e. good enough to merit Cy Young consideration.
So here are my Top 10 for the NL Cy Young:

  1. Roger Clemens
  2. Andy Pettitte
  3. Dontrelle Willis
  4. Roy Oswalt
  5. Chris Carpenter
  6. Pedro Martinez
  7. Jake Peavy
  8. John Smoltz
  9. Carlos Zambrano
  10. Brandon Webb

My AL Cy Young argument is what you just read above: ignore Colon's wins, because they mask the fact that he's just not that good a pitcher. Here are the leading candidates and their essential stats:

Bartolo Colon (21-8, 3.48, 222.2 IP, 26 HR, 43 BB/157 K)

Johan Santana (16-7, 2.87, 231.2 IP, 22 HR, 45 BB/238 K)

Mark Buehrle (16-8, 3.12, 236.2 IP, 20 HR, 40 BB/149 K)

Mariano Rivera (7-4, 1.38 ERA, 43/47 SV, 78.1 ip, 2 HR, 18 BB/80 K)

Yeah, once you get past Colon's win, you see that he can't hold a candle to Johan. And after hearing my rant about ERA, I'm sure you can guess that Johan is my choice. Other than the wins, he's inarguably the best starting pitcher in the league (it ain't his fault the Twins can't hit).

But what about Rivera? This is a tough question. How can you compare a closer to a starting pitcher? Sabermetricians aren't exactly positive, but they've got a much better idea than mainstream analysts. Really, mainstream baseball commentators have no clue how to compare a good starter to a good reliever. Just look at Rollie Fingers and Dennis Eckersley getting the Cy Young. I'm not even going to mention Mark Davis and Steve Bedrosian ...

In my opinion, Rivera has been good -- very good. But he just can't trump Johan. If you're going to pitch 155 fewer innings (albeit more important innings), they have to be a huge leap better than the starters. I pick Rivera as my #3. Win Shares generally agrees, giving Rivera 17 WS against 23 for Santana (although they give 21 to Jon Garland for no apparent reason, so take from that what you will). If you've got the RSAA numbers or VORP numbers, feel free to send them in. The subscription to baseball prospectus is not exactly within my budget (I'm an actor), so Win Shares are the only free statistics I've got.

Here are my AL Top 10 in the Cy Young Award voting:

  1. Johan Santana
  2. Mark Buehrle
  3. Mariano Rivera
  4. Kevin Millwood
  5. Bartolo Colon
  6. Jon Garland
  7. Roy Halladay
  8. Carlos Silva
  9. Kenny Rogers
  10. John Lackey

Just you wait for my Rookie of the Year and other random awards analyses ...

No comments: