Covering the Front and Back Pages of the Newspaper
July 8, 2003
BASEBALL: 2003 Mid-Year NL East DIPS Report
One of the major questions that comes up by this point in the season is: who's for real? It's late enough, just past the halfway mark, that we are apt to start giving some weight to this year's stats. But as we well know by now thanks to Voros McCracken, pitcher stats can be heavily influenced by luck and defense on balls in play. Thus, a pitcher who's success is dependent on those factors may be a bad bet to keep it up.
So, I decided to use Voros' rough in-season Defense-Independent Pitching Stats (DIPS ERA) calculations* to take a look at how a few interesting pitchers stack up when you take balls in play out of the equation.
* - Scroll down to the bottom of this post for details on the formula; the formula I used may need to be adjusted downward a bit, given that it produced a DIPS ERA of 4.50 for the National League as a whole compared to a league ERA of 4.32. The AL numbers were closer, with a DIPS ERA of 4.55 for the AL compared to a league ERA of 4.54.
I'll run a few of these here and do more later this week if time permits. Let's start with the NL East (all stats through Monday night's action):
For the Mets, I'll just run the four starters who have significant innings, plus the Mets' "All-Star" closer, Armando Benitez:
For the most part, the DIPS formula actually suggests that the Mets' primary pitchers have been even worse than it appears, especially Glavine, Trachsel and Benitez. This seems odd, since the Mets' defense is one of the worst in the National League at turning balls in play into outs, and David Pinto's adjusted calculations (taking extra base hits into account) also have them near the bottom.
The good news: Jae Seo is for real, with just 20 walks allowed and 7 home runs in 102.2 IP. Seo is a rarity, a "Tommy John-type" pitcher (low BB, low K, low HR) who's not lefthanded and not a sinkerballer (in fact, these days, most control specialists are big fly ball guys like Brad Radke and Rick Reed).
On to the rest of the division:
For the Marlins, let's look at the vaunted young arms in the rotation:
Main lesson: the Marlins' most and least successful starters are closer together than they appear. Of course, this doesn't take account of park effects, which may be helping Dontrelle Willis in particular. And Willis may not be a legit 2.13 ERA guy (who is?), but he's been plenty effective nonetheless.
DIPS appears once again as The Great Leveler: it turns out that Horacio Ramirez and Russ Ortiz haven't really pitched that much better than even the struggling Shane Reynolds. And yes, John Smoltz is amazing, but nobody's that good.
DIPS is pretty down on the Expos pitchers (or up on the Expos' defense, depending how you look at it): basically everyone on the staff either allows too many home runs or doesn't strike anyone out. The exception is improbable closer Rocky Biddle, who's getting the job done with whiffs and just 2 homers allowed in 42.1 IP, thus allowing him to post respectable numbers despite a ghastly 5.31 BB/9IP. Advice to Rotisserie owners of Claudio Vargas: sell.
With the Phils having the most-efficient defense in the NL, you'd expect DIPS to downgrade some of their pitchers. Of course, none of that will matter much if the defense plays like this all year, but Millwood's status as staff ace should become clearer as the year wears on (and Millwood's defensive support looks even worse when you consider that one of his starts was a no-hitter), while Randy Wolf's breakout year looks a lot shakier when you consider that more than 1/6 of his hits allowed this year have left the yard, and Brett Myers is likewise way over his head.
* - Unfortunately, there seems to be some confusion over McCracken's formula; it appears that, as of 2001, the formula was as follows:
DIPS ERA = ((IP*2.4)+(H*.83)+(HR*11.05)+(BB*2.81)-(SO*1.59)) divided by ((IP*0.71)+((H-HR)*.244)+(SO*.097))
This is the corrected version at the end of the comments section in his Baseball Primer piece, although admittedly it still doesn't give me the results Voros reported for 2001 -- it consistently gives results around 0.15 higher. So, perhaps the DIPS ERAs I'm reporting here will be a bit higher than they should, and as noted above, the NL DIPS ERA as a whole is out of whack with the league ERA. If anyone's aware of an improved version published since then, let me know.