Longtime readers know that one of my pet peaves is the arbitrary designation of how good a starting pitcher is. People like to classify guys as “#1 starter”, “#3 starter” etc. Its not an uncommon way to classify players in sports – because its fairly intuitive. I’ve been involved in a lot of debates about whether or not Scott Gomez was a “true #1 center’ for the Devils – and I absolutely hate the designation.

Each team in a normal rotation has to carry 5 starters. The Yankees have Javy Vazquez, Andy Pettitte, C.C. Sabathia, and A.J. Burnett, and the last spot is up in the air. Besides for an extra start or two possibly handed to C.C. Sabathia due to off days, and a few taken away from the Joba/Hughes competition for the same reason, but for the most part until the playoffs the rotation order is completely meaningless. Whether or not Javy Vazquez is the #2 or #4 starter has absolutely no impact on the game. But the label has a lot of intuitive appeal, which in my opinion is the most important part of a statistic, so I think that giving the language some real meaning, we can better place in our minds our players.

I am a fan of a methodology used by The Hardball Times awhile back. To quote the article (using the 2006 Twins):

For the purposes of this article, it’s necessary to define exactly what a #1 starter (or #2, or #3) is. To keep things as simple as possible, I used ERA as a measure of pitching ability. I also figured that each rotation spot accounts for 32 starts. On many teams, the #1 guy isn’t the same for the whole season. For example, let’s look at the 2006 Twins. Here are all of the pitchers who made more than one start for Minnesota last year:

Starter GS      ERA
Liriano 16      2.16
Santana 34      2.77
Bonser  18      4.22
Radke   28      4.32
Garza   9       5.76
Silva   31      5.94
Baker   16      6.37
Lohse   8       7.07

By ERA, Francisco Liriano was the best of these guys, but he only made 16 starts. So, he made half of the “#1 starter” starts. Since Johan Santana is next in line, I assigned 16 of his starts to round out a composite #1 starter. Thus, the Twins #1 starter was half Santana, half Liriano. Santana’s remaining 18 starts were assigned to the composite #2 starter.

Intuitively speaking, that distribution is a reflection of the fact that, while Liriano was in the rotation, Santana was #2. When Liriano was in the bullpen or on the disabled list, Santana was #1. Here’s how that shakes out for the Twins staff:

Starter  GS      ERA
Liriano  16      2.16
Santana  16      2.77
#1 Total 32      2.47

Santana  18      2.77
Bonser   14      4.22
#2 Total 32      3.40

Bonser   4       4.22
Radke    28      4.32
#3 Total 32      4.31

Garza    9       5.76
Silva    23      5.94
#4 Total 32      5.89

Silva    8       5.94
Baker    16      6.37
Lohse    8       7.07
#5 Total 32      6.88

I think its a pretty good method. In a perfect world, we would use ERA+ or something more complex, but for the purposes of keeping it intuitive, I’ll stick with the THT method. The biggest problem with the method is it doesn’t account for innings, but it is meant for the purpose of illustration more than calculation. The article crunches the numbers for all 2006 AL teams. I wish I had 2009 numbers, but I don’t really have the database power to generate those numbers. From 2006:

• #1 Starter 3.7
• #2 Starter 4.24
• #3 Starter 4.58
• #4 Starter 5.09
• #5 Starter 6.22

If we look at the 2009 Yankees, we get the following:

• #1 Starter: 3.37 ERA
• #2 Starter 4.00 ERA
• #3 Starter 4.15 ERA
• #4 Starter 4.70 ERA
• #5 Starter 7.5 ERA

Relative to 2006, the Yankees held a pretty solid advantage in the #1-4 spots, and completely fell apart at the #5 spot. Chien-Ming Wang and Sergio Mitre are to blame. I think that this really shows what the addition of Javier Vazquez means to the team. Even if Chamberhughes doesn’t improve on Joba’s 2009 performance, the Yankees essentially get to replace 32 starts of 7.5 ERA with 32 starts of Javier Vazquez, who at the very least is a pretty good bet to have an ERA in the low 4s. If Vazquez were to replace 6 innings a game with an ERA of 4.20, my quick and dirty calculation says that is a 6 win improvement over last season.

Injuries and spot starts will happen, so this isn’t a straight-up replacement, but I think it is a very good way to conceptualize why the 2010 Yankee rotation is built so well.

Furthermore, I think that this method demonstrates how valuable healthy starting pitchers are, even if they don’t excel. If a team starts the year with 5 starters who each get 5 starts, even if the worst ones aren’t very good, they have a pretty big competitive advantage. The slope is quite steep for teams once they start to reach down in to their depth charts. This is where the Yankees grade quite well – their top 4 starters have been exceptionally healthy in the past few years. In fact, they average 33.25 starts per season in the past 2 seasons.

And to make one last point: what if some combination of Chamberhughes becomes a pretty good pitcher? They both certainly have the ability to throw up an  ERA in the 3s. If they were to, for instance, toss in 32 starts with a 3.80 ERA and 6 innings per start, and you combine that with Javy’s addition, the rotation could be 8 wins better than last year. And the Yankees won 103 games in 2009.

Long story short: current roster construction combined with some good health luck could mean a rotation capable of 1998-like heights.

In part 2 (hopefully some time this week, real life is kicking my butt right now) I’ll take a look at what really interests me about this: evaluating starting pitching prospects. So I’ll basically be bashing BA, which is always fun.