Regression Will Find You

As we head into the all-star break, the pause gives us a chance to look at some players who have had a few months worth of performance that we just don’t think they can sustain – in short, they’re poised to regress to the mean, for one reason or another. For the second half of 2008, here’s the pitching staff for the All Regression Squad:

Justin Duchscherer: .216 BABIP, 4.5% HR/FB.

Duke’s first half is a good story, and he’s pitched very well at times, but there’s no way those two numbers are sustainable. He won’t implode, but he’s not going to keep his ERA under 2.00 all year.

Gavin Floyd: 3.69 ERA, 5.09 FIP.

Good luck keeping that BABIP at .226, and when it inevitably rises, so too will the curtain on what artificially appears to be Floyd’s breakout year.

Dana Eveland: 3.7% HR/FB rate.

I don’t care how big your home park is, that just can’t last. With his command problems, the depressed home run rate is really the driving force behind his success – when that goes, he’s going to be in trouble.

Jake Peavy: 83.7% LOB%.

Peavy is awesome, but he can’t keep stranding runners like this. No one can. He’s still going to be very good, but not quite this good.

Scott Olsen: 3.77 ERA, 5.00 FIP.

His velocity is gone, his strikeout rate is gone, his groundball rate is gone. It’s all smoke and mirrors at this point, and there’s just no way he keeps getting people out with the repertoire he’s currently throwing.

Print This Post

Dave is a co-founder of and contributes to the Wall Street Journal.

11 Responses to “Regression Will Find You”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Chuck says:

    What about Ryan Dumpster? A .258 BABIP is not sustainable for a player with career .308 average and a LD% that is in line with his career average.

    Vote -1 Vote +1

  2. beamsplitter says:

    Peavy may regress, but your choice of LOB% as support is a strange one.

    You say no one can strand runners at 83.7% but Peavy did it over 166 innings in 2004. Also, between ’04 – ’07 he’s been at least in the upper 70%s. So, although his strand rate today is arguably a bit high, it certainly isn’t unsustainable.

    Now that I look at it, Peavy’s 2008 so far is remarkably similar to his 2004… I’ll take the under on this particular prediction, Dave ; )

    Vote -1 Vote +1

  3. Dave Cameron says:

    And after Peavy’s flukey LOB% in 2004, it regressed quickly in 2005.

    Check out the list of players who have posted a LOB% of 80% or higher from 2004 to 2007, and then check their following years LOB%. You’ll see regression across the board.

    No one sustains an 80% or higher LOB%.

    Vote -1 Vote +1

  4. Sky says:

    Many of the stats that point to flukey pitchers will also result in higher than normal LOB rates, because the fewer baserunners you allow and the fewer homeruns you allow, the lower the percentage of baserunners that score. (The first runner in an inning rarely scores, but adding a increases the first’s chances of scoring and also could score himself.) Good pitchers tend to have good LOB%s not because it’s a skill, but because better pitchers allow fewer baserunners and homeruns.

    So when I see a high or LOB% rate, I often first look to a weird BABIP or HR/FB. Then, if those are in line, I chalk it up to the LOB%, like in Peavy’s case. His career rate is 77%, which is probably fine given his talent, PETCO, and the Padres’ defense.

    Vote -1 Vote +1

  5. chronosaurus says:

    Not a single mention for Cliff Lee?

    Vote -1 Vote +1

  6. Brian says:

    Cliff Lee’s ERA is real in that it perfectly reflects what his raw numbers would suggest: his .298 BABIP is on par with the league norm and his 78% strand rate sounds about right for a guy who has only allowed 27% of batters to reach base and very few home runs. It is not real in the sense that he is almost certainly not this talented and any regression that he sees will come in the form of a decline in controllable skills. His K% has soared to an all-time high while his BB% has dipped to an all-time time low, which could very well be legit, but it’s just as likely that it’s not. His ridiculous home run rate (one every 98 PA!) figures to regress as his super low 4.1% HR/FB does, but at least his newfound groundball ability—and increased punchout rate, if sustainable—will mitigate the bulk of the damage.

    Vote -1 Vote +1

  7. Mike says:

    Worth noting re: Eveland:
    Eveland yielded 8 HRs over 246 innings from 2005-2007 in the minors, so it’s reasonable to think that his depression of home runs is at least to some extent real.

    On the other hand, his command is terrible and even a slight uptick in the HR-rate could spell disaster.

    Vote -1 Vote +1

  8. Bill says:

    What about Dice-K? Here’s the comparison between Dice-K and Daniel Cabrera (leading into their H2H matchup right before the break) so far this year:

    D.Cabrera 69K 51BB 1.37WHIP
    Dice-K 70K 52BB 1.37WHIP

    And some other numbers:

    D.Cabrera 6-4 4.40ERA 124.2IP
    Dice-K 9-1 2.84ERA 82.1IP

    Funny aren’t they?

    Obviously luck and run support is playing a huge part in Matsuzaka’s 10-1 record. Dice-K’s FIP is 4.10 and Cabrera’s is 4.95, so there’s definitely a gap there in their true performance, but it’s nowhere as wide as their W-L records and ERA indicate (thanks to diverging luck and run support they’re experiencing and receiving so far this season).

    As a Red Sox fan, I can’t tell you how many times I’ve somehow seen him walk a couple of guys, hit a batter, and give up a hit or two, all in the same inning, and not give up a run. No way Dice-K can keep this up.

    Vote -1 Vote +1

  9. Jay says:

    I’m guessing zero since it’s extremely rare for four batters to reach base without scoring.

    Vote -1 Vote +1

  10. Brian says:

    RE: Peavy

    But the topic is- who’s numbers will regress in the second half. Not whose numbers could quickly regress the following season.

    Vote -1 Vote +1

  11. Brian says:

    This is considered analysis. Duke won’t keep his ERA under 2, wow that is some in-depth stuff.

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>