Sunday, December 05, 2004

Rating the Free Agent Starters

Here's a little free homework for Jim Bowden on this year's free agent class. I compiled two statistics for each of the major free agent pitchers that are out there (throwing in Kris recent signee Benson): ERA + in 2004 and BABIP in 2004.

ERA+ is defined by Baseball-Reference.com as "the ratio of the league's ERA (adjusted to the pitcher's ballpark) to that of the pitcher. [Greater than] 100 is above average and [less than] 100 is below average. lgERA / ERA" The all-time leader in career ERA+ is Pedro Martinez, who, based on that statistic, it could be argued is the best pitcher of all time (his 1st place career ERA+ of 167 is a mind-boggling 13% better than the next best, Lefty Grove. Pedro will get worse as he gets older...but I digress).

BABIP is a measure of the Batting Average allowed by that pitcher when balls are put into play. As explained by Voros McCracken, a pitcher has little control over what happens to a ball once it is put in play. A pitcher only really has control of the situation when he walks a hitter, strikes the hitter out, hits the hitter with a pitcher, or allows a home run (fielders can't screw that up). McCracken proved that, in the aggregate, a low BABIP in one season doesn't translate into low BABIPs in other seasons, all other things being equal. Pitchers simply don't have much control over how many balls their fielders turn into outs. Thus, a very low BABIP means a pitcher has gotten especially lucky with the play behind him. A very high BABIP means the opposite.

Without further ado, here is the 2005 starting pitcher free-agent class, ranked according to how I think they performed in 2004:

PitcherERA+BABIP
Brad Radke1360.293
Jaret Wright1310.292
Carl Pavano1370.282
Pedro Martinez1250.292
Matt Clement1230.279
Odalis Perez 127 0.262
Al Leiter 133 0.240
Derek Lowe90 0.342
Woody Williams 100 0.289
David Wells 108 0.273
Kris Benson 97 0.294
Aaron Sele91 0.313
Esteban Loaiza 84 0.311
Shawn Estes 86 0.301
Ismael Valdez 78 0.282


I've color-coded the BABIPs - green means that the pitcher was likely relatively unlucky last year, the blue means that their BABIPs were in line with the league average, gold means that they may have been a bit lucky last year (a warning sign), and red means that they were most definitely lucky last year.

And if you think the Mets may have been damn foolish for signing Kris Benson to a $7.25 million per year deal, well, you should take that up with Mets management. I don't feel like getting into an argument with you. However, signing Kris Benson apparently has benefits for his teammates.

It's no surprise that Radke, Pavano, Wright and Martinez are the cream of the free agent crop. But Radke and Wright both seem to be better values than Pavano and Martinez based on their 2004 performance. Either one would be a terrific signing for Bowden provided that he had the budget for it. And I remain high on Matt Clement, but he does seem to be a rung down from the top starters.

A couple of points about those I would stay away from. Odalis Perez got very lucky in a forgiving Dodger Stadium in 2004. Al Leiter got extremely lucky in a very forgiving Shea Stadium in 2004. Whatever temptations a GM may have, I would think long and hard before giving either of them significant deals.

Finally, Derek Lowe simply is not as bad as he pitched last year. His .342 BABIP was absurdly unlucky, and he's likely a guy who could come back with an ERA 1 run lower next year just through getting average defensive support behind him. He does not give up many homers, and would likely pitch well in RFK stadium. At the right price, Derek Lowe could be a big benefit to the 2005 DC-9.

7 Comments:

At 11:27 PM, Blogger DM said...

Very interesting, SNV. For a second I thought you were being hard on Al Leiter, then I looked at his 4-plus walks per nine innings last two years, and agree he is radioactive. Ideally he would retire and become the Nats color commentator, because he is GREAT at that.

 
At 12:24 AM, Anonymous Anonymous said...

Isn't there also a sabermetric that measure how many batted balls were line drives v. fly balls v. ground balls? It seems that for a pitcher to be very unlucky he would have to give up a lot of ground balls, a high percentage of which are hits. A lucky pitcher would be one who gave up lots of line drives but has a low BABIP. But a pitcher who gives up lots of ground balls and has a low BABIP might be someone who fools a lot of hitters with off speed stuff or changes locations very well. Like Greg Maddux. I do believe the line drive stat is out there so it'd be interesting to measure BABIP against hit type to see who is really lucky and who just is a ground-ball pitcher.

 
At 10:04 AM, Blogger SuperNoVa said...

Ah, Mr. Anonymous, if that's your real name, you've touched on one of the aspects of BABIP that has subtlety. In fact, McCracken and others later showed that pitchers have some (but not much) control over their BABIP. It turns out that fly ball pitchers do have a bit of an advantage in BABIP over ground ball pitchers. But, then again, fly ball pitchers also tend to strike more people out and give up more homers, so it's something of a wash.

I use BABIP only to see whether a pitcher is very lucky - has a BABIP below .270 or so, 10% below the norm - or very unlucky. Can you say that a pitcher got lucky because his BABIP was .290 and the league average was .295? Nope. But as the pitcher strays from the average, the questions rise.

For example, Johan Santana, who had a wonderful year this year, had a BABIP of .250, which is a little more than 15% lower than the AL average (.296). If he had a league-average .296 BABIP, it's likely that he would have had an ERA 0.3 or so higher - maybe enough to give Schilling the Cy Young award. And although Santana is a terrific pitcher, don't expect him to repeat his 2004 performance next year assuming the same peripherals (walks, homers, strikeouts). He should slip a little bit (but still be a very good pitcher).

 
At 10:07 AM, Blogger SuperNoVa said...

DM, I think Al Leiter is a terrific commentator - probably the most informative one I've listened to in a while - and agree that he would be great in the Nats booth when his playing days are over.

 
At 10:13 AM, Blogger Harper said...

Other than Pedro, what are they like for their career and what are their ages? Radke and Wright might be better values, but Wright could be a one year wonder and Radke is what, 50? I could hit off him. He sucks!

Any pitching would help, but this team needs a definitive #1 starter more than anything else and neither Radke or Wright fits this bill. Unless we can get a great deal on one of them, I say hold off, save your money and see if we can't get a real #1 next year.

 
At 11:43 AM, Blogger DM said...

Dexys comment is a good one, and essentially the reason I looked into Al Leiter's stats, thinking that the ERA+ may not be the best stat to use here. What I found was not good for Al -- his BBper9 is going up and SOper9 is going down over the past two years.

This is probably an obvious question, but is there a stat that combines the things that pitchers can control, e.g. that combines BB, SO, and HR into a reference number? I couldn't find it quickly last night. That seems to be the number you want to look at more closely than ERA+.

 
At 12:04 PM, Blogger SuperNoVa said...

Holy beejeezus, DM, you've created a new stat! Combining BB/9, SO/9 and HR into one easy-reference stat would be fantastic.

One could combine K/SO with K/9 (say 3 * 9 = 27), which would provide a meaningful indicator. Someone with a 9 K/9 but with a 2 K/BB would be an 18 and someone with a 6 K/9 with a 3 K/BB would also be an 18.

Question - is a pitcher with a 9 K/9 and a 2 K/BB more or less effective than a pitcher with a 6 K/9 and a 3 K/BB rate? It's probably a pretty close call, we'll have to check.

Now, how do you combined that with HR rate? What if you divided by HR/9? Someone with a .5 HR/9 (e.g., 10 HR in 180 innings) and a 6 K/9 and a 3 K/BB would be a 36.

The thing I worry about that is that it would overweight home run allowance.

I think you could probably play with this and come up with a linear regression to see which stat has the tightest fit to runs allowed over the long run. You'd almost have to do it on a league-wide basis for several years to figure it out.

 

Post a Comment

<< Home