Tuesday, February 16, 2010

Spurned by SIERA

Today, we finish the Cubs-SIERA analysis, taking a look at the bottom of SIERA's barrel. But, before we dive into that, I'd like to mention mb21's ACB post about Carlos Silva. I am beginning to campaign for Silva's forgiveness because: a) I really do hope he turns his career around and b) I may have looked at his numbers and scoffed him too quickly. That being said, I really would be surprised if he has a good season. I suspect he gets designated for assignment by mid-season. :(

Anyway, Cubs Stats readers may recall my previous post, which both explained SIERA and applied it to the 2009 Cubs. In that post, we discovered SIERA dislikes these pitchers:

The Kind-of Losers (anyone with a higher SIERA than xFIP)
Sean Marshall: (SIERA) 4.01, (xFIP) 3.82
Randy Wells: 4.34, 4.24

Also:
Carlos Silva: 5.65, 5.38

Now, what I found most intriguing about the differences in SIERA and xFIP is that the maximum gain from SIERA was around 0.94 (Russ Springer) and the maximum loss was equally close to 1 (J.J. Putz). In other words, there is really not a whole lot of variation there. But, considering SIERA uses a lot of similar inputs as xFIP, we should expect that any variation is pretty close to being significant (even if it is not statistically significant, which is a whole different can of worms).

So what are the profiles of these pitchers that SIERA would punish them? Or, more importantly, do any of these pitchers have profiles more akin to recieving SIERA's blessing rather than spurning?
Sean Marshall earned some above-average ground ball rates (GB%), but his strikeouts and walks were close to league average (K/BB 2.13; league average is about 2.00, and higher is better). Also, he had a normal home run rate. Therefore, SIERA doesn't see anything to reward.

Randy Wells had a great rookie season, but was also pretty lucky. Almost every metric tells us that. He had an unusually low HR-rate (0.76 per 9 IP; league average is about 1.00) with fewer strikeout than you would like for a guy giving up a standard amount of flyballs. In other words, SIERA can see through the luck, too.

Carlos Silva really only pitched a handful of games in 2009, a total of 30 IPs. It's hard to critique on that small of a sample, but SIERA rightly detects that: a) he struck no one out and b) surrendered long balls like mad.
Ultimately, it again appears that SIERA has done a pretty good job identifying what the marginal statistics confirm. With the exception of Carlos Marmol, I can't think of any SIERA results that left me scratching my head.

3 comments:

  1. I haven't had enough time to read through all of the BP articles yet. I've read the first 3 parts and then started reading around about what others had to say. I'm not too confident in making judgments yet, but that Marmol SIERA is odd.

    ReplyDelete
  2. Brad, do you happen to know the formula for tRA? I'm trying to take your SIERA calculator and add FIP, xFIP and tRA to it. It's not like FIP is that hard, but maybe some others would use it if they could easily calculate it. Also, it would be nice to have all of the advanced pitching metrics available on one spreadsheet. I can't find the formula for tRA though.

    ReplyDelete
  3. Gee, I've looked for tRA (and tERA) in the past and found nothing. Maybe I can ask around the blogs. I've got some street-cred with the guys at DRaysBay, and they know a lot of this stuff. I'll let you know what I find.

    ReplyDelete