Wednesday, March 02, 2011

Duensing's Just Deserts

A few weeks ago, I peeked back at Twins pitching stats. One name kept coming up, over and over:

But this analysis has really opened my eyes on the year [Brian] Duensing had. He was among the leaders in ERA, quality starts, eating innings and WHIP. For two years now we've seen what he can do for a half year in the rotation. I think it might be time to see what he can do for a full year.

Over those two years, Duensing has started 22 games and tallied 138.1 innings, which is about four months of work. The statistic that hopeful Twins fans will latch onto is his miniscule 2.93 ERA. Some of his other stats are more mediocre – 136 hits, 84 strikeouts, 37 walks and 12 home runs. Those are all about in the range of what you would expect for a slightly above-average starting pitcher, not an outstanding pitcher.

There is a statistic that can explain some of that paradox. It’s called BABIP, as in Batting Average on Balls In Play. Duensing’s was .284 as a starter, which is also above average, but suggests he was a little lucky – but did you really need me to tell you that? You’re talking about a guy without great stuff, who has a career ERA in the minors of 3.61. Do you think he’s a 2.93 ERA pitcher? We just overlooked a Pedro Martinez in our farm system?

Of course not. The question is – is he close to that?

To that we’ll go to yet one more statistic called FIP (for Fielding Independent Pitching). FIP is calculated ignoring those pesky batted ball, instead using several of those “slightly above average” stats to determine a number that looks like an ERA. According to the excellent site, Duensing’s was 3.91 last year as a starter. This is significant because FIP is a little bit better predictor of a future ERA than ERA is.

So no, he’s not Pedro Martinez, or even particularly close. But if Ron Gardenhire, who announced yesterday that Duensing already has secured a spot in the rotation, gets a 3.91 ERA for a full year out of the southpaw, he’ll need to up his dosage of cherry extract to handle the backflips he’ll be doing.

Not only do the statistics support Gardy’s decision, so does common sense. Gardenhire needs to manage this team and manage this pitching staff, and one thing that managers universally want to drive home to their team is that performance matters. For two years, Duensing has performed, posting quality start after quality start, as well as accepting whatever role the team needed. It’s entirely reasonable that he get his just deserts a little early this spring.

Late addition: A friend pointed out to me that the initial version of this entry was incorrectly spelling the term "just deserts" as "just desserts." According to this wiktionary entry, "deserts" is the plural of "desert," which is "that which one deserves." Just a late note for you future writers out there....


Anonymous said...

You said FIP is a better predictor going forward than ERA. Has that actually been shown to be true or is it just assumed to be true (b/c it should be)?

Anonymous said...

"This is significant because FIP is a little bit better predictor of a future ERA than ERA is."

And xFIP and SIERRA are better regressions than FIP. They paint a bleaker picture.

"Not only do the statistics support Gardy’s decision, so does common sense."

Do the numbers support your claim, or did you cherry pick numbers that did? xfip and sierra both over 4, he has mediocre control paired with a horrendous strikeout rate, a hr rate below expectation and a dp rate above expectation. Duensing has incredible room for regression. But at least FIP thinks a likely sub 4 era candidate.

John said...

Actually, I did a study that showed FIP is a better predictor than ERA. You can find it here:

I didn't use xFIP because I don't know how good a predictor it is, and I didn't use SIERRA because I don't know anything about it. I don't know what you mean by "better regressions." Do you mean better predictors?

Anonymous said...

Brian Duensings 2010 Siera was 4.52.

Anonymous said...

The stat is SIERA not SIERRA. My mistake. It stands for skill interactive earned run average. Baseball prospectus uses it and I believe the developed it. It uses a more robust fit of gb,k and bb rates to past run scoring data. Here is BP's introduction to the stat if youre interested.

Anonymous said...

desserts, man.