Rotational Metrics Revisited
Reader-Submitted Suggestions/Corrections
1/26/04
Jason Moyer

Last week's article comparing some of the better starting rotations going into 2004 prompted a bit more feedback than I had anticipated. Quite frankly, I dropped the ball on a few details, and several readers were kind enough to correct me. This week, I'd like to address some of the more interesting email I've received on the subject.

Estes is off the Cubs' roster. What would happen if you ran your analysis with somebody who might become the Cubs' fifth starter like Cruz, Guzman, or Maddux? I'd be curious to find out. -- Jay Gainer

In my initial study, I had mistakenly assumed that the roster information I was completely up to date. As such, the data I was using for the Cubs, Red Sox, and Astros was outdated.

In the case of the Red Sox, the spot previously held by John Burkett will likely by taken by Byung-Hyun Kim. Since most of his innings came in relief last season, I'm not entirely certain how to deal with him in this study, so take my findings here with a grain of salt. I'll use his win shares earned in Arizona, where he was a full time starter, for this analysis. He was a bit more effective, inning-for-inning, as a reliever in Boston, but I'd like to disregard his stint as a reliever if possible.

Likewise, it is now clear that Robertson will not be the Astros #5 this year, as Jimi Williams has assigned that role to Tim Redding.

The Athletics and Phillies were, to the best of my knowledge, accurate, so I will use the same rotations for this study. For the Cubs, since it seems increasingly likely that Maddux will be returning to Chicago, and given the lack of good starting data for Cruz, I'll pencil Maddux into the final spot in the rotation.

One other major point that I missed last week, and then I'll move on to the study proper:

Your article is interesting. However, you seem to have made a few decisions that - intentionally or otherwise - slant the final product.

Primarily, I'm talking about your decision not to list the Yankees because they don't have "a solid idea of what their rotation is going to look like." That's just poppycock.

Mussina, Vazquez, Brown, Contreras, and Lieber seems pretty clear to me and most other observers. You may want to argue the point that Lieber isn't a lock, but he's got a better shot of being in the rotation for the Yanks than Robertson does in Houston or Estes (listing a free agent as the team's number five is another specious decision) does in Chicago.

The Yankees UNQUESTIONED top-four starters had a raw Win Shares total of 67, out pointing the Phillies, Astros, Marlins, and Red Sox even WITHOUT the Win Shares contributed by a number five starter (I don't know Lieber's 2002 totals).

If we use weighted Win Shares, the Yankees' rotation totals 146. Again, without the Win Shares of a number five starter. If we pencil in Lieber for 6 Win Shares (a reasonable assumption for a former 20-game winner AND reasonable considering the 9 Win Shares you allocated to Milton and the 14 you allocated to Burnett coming off Tommy John surgery), the Yankees' weighted Win Shares total of 176 surpasses every single team in your study.

Finally, when looking at Win Shares over average, the Yankees come in at a whopping 28.5, even without a fifth starter. In short, Lieber could be -8.5 and the Yankees would still project as having the best rotation.

I'll admit that the math here may be slightly off. But I believe the gist of my assertion is correct. Perhaps I'd feel differently if you had excluded the Marlins, Phillies, and Astros - three teams with as much or more rotation uncertainty that the Yankees. But speculating on those rotations while intentionally deciding to omit the Yanks did both the Yanks and your article a disservice. -- Michael Beckenbach

Thank you Michael for both the information on the Yankees rotation (which I should have had) and for running my initial study for them. Clearly, it was a huge oversight on my part to leave the Yankees out of my comparison.

Since I already did a bit of a win shares analysis last week, sticking entirely to 2003 win shares and win shares above average, I'm going to alter my approach slightly. This week I'd like to use 3 years of data, where possible, including the standard pitching stats (innings, strikeouts, walks, and homeruns) and total win shares.

For the sake of showing how I'm working with this information, here are the 2004 Chicago Cubs. For Zambrano, I'm limiting the data to 2003, as it's the first season he served as a full-time starter.

Cubs	GS	IP	K	BB	HR	WS
Prior	49	328.0	392	88	29	29
Wood	93	599.0	700	289	62	43
Zambrano	32	214.0	168	94	9	18
Clement	95	576.0	520	249	55	15
Maddux	104	650.7	415	105	58	50

Assuming (perhaps incorrectly) that each starter receives the same number of starts, we can use basic algebra to create an estimated performance for each pitcher over the course of a single season. I realize some of these pitchers will end up with more than 32 starts and some will have fewer. Who will have more/less would be guesswork at this point, and since much of that will be determined by injury and manager usage, I'd rather not speculate.

Cubs	GS	IP	K	BB	HR	WS
Prior	32	214	256	57	19	19
Wood	32	206	241	99	21	15
Zambrano	32	214	168	94	9	18
Clement	32	194	175	84	19	5
Maddux	32	200	128	32	18	15
TOTALS		1028	968	366	86	72

Here are the totals for each team using this method:

Team	IP	K	BB	HR	WS
Cubs	1028	968	366	86	72
Red Sox	1042	958	269	101	86
Athletics	1040	771	347	83	78
Phillies	1003	731	310	115	62
Astros	997	868	308	99	71
Marlins	977	792	361	93	58
Yankees	1050	936	269	101	84

The Astros fare much better this week, presumably because the use of 3 seasons of data makes up for the win shares they lost due to injuries last year. The Red Sox, Yankees, and Athletics are the cream of the crop, and the Phillies and Marlins lag behind.

On a final note, would you believe that Kevin Millwood has averaged just 12 win shares per 32 starts over the last 3 years? And that the Phillies are so desperate for pitching that they'd rather pay him 8 figures than bring up a prospect who would likely put up similar numbers?

Send in your comments and questions!