Giving a Questionable Start to Baseball’s Best Starter

There was speculation that, had the Dodgers lost Game 3 to the Braves, they would’ve asked Clayton Kershaw to start Game 4 on short rest in order to help the team avoid elimination. The Dodgers wound up clobbering the Braves in Game 3, moving a win away from the NLCS, and now they’ve asked Clayton Kershaw to start Game 4 on short rest anyway. The official announcement was made earlier Monday, with Kershaw figuratively taking the ball from would-be starter Ricky Nolasco.

Generally, something like this is a desperation decision, but clearly the Dodgers are a team that’s not desperate. They’ve got two chances to get rid of Atlanta, and they were guaranteed to have Kershaw get one of them. Now they’re lined up to give the starts to Kershaw and Zack Greinke, instead of Nolasco and Kershaw. That is, if a Game 5 is necessary. In making this decision, the Dodgers have demonstrated that they don’t want to see a Game 5 at all.

It’s surprising, for the two reasons: the Dodgers are ahead in the series, and Kershaw has never before started on short rest. But just because something is surprising doesn’t mean it’s wrong, so we might as well examine right quick. And the math breaks down kind of like this: how much better is Greinke than Nolasco? How much worse is short-rest Kershaw than regular-rest Kershaw?

Nolasco contributed three ugly starts for the Dodgers in September. They’re his most recent three starts, over which he allowed 19 runs in 12 innings. In Greinke’s September, he allowed seven runs in six starts. It’s evident who finished better, but it’s also evident that you can’t just make decisions on September statistics, and in a lot of ways Nolasco was Greinke’s equal over the course of the year.

The walks were more or less the same. The strikeouts were more or less the same. Nolasco allowed four more home runs, but he also threw 22 more innings. He’s got himself a long history of not quite pitching to his peripherals, but Greinke does, too. Unscientifically, I think it’s safe to say that Greinke is the better starter, especially right now, but we’re talking a matter of degrees. It’s hardly a big difference, although I suppose come playoff time you want to maximize all your percentages.

Really at issue is Kershaw on short rest. We can’t compare him to himself, because he’s never done this before. We do have a history of other starters going on short rest in the playoffs, so that’s going to have to do. Below, you’ll see statistics for two groups, in postseason play between 1995-2013. The first group is starters who went on three days’ rest. The second group is starters who didn’t. Understand that the first group should be selective for better arms, because you don’t put mediocre starters on short rest in October, or ever. Typically, it’s aces who go on three days’ rest, and the numbers are telling:

Short-rest group: 4.66 ERA, 5.13 RA
Other group: 3.99 ERA, 4.30 RA

On short rest, starters have faced fewer batters, on average. They’ve posted worse game scores, on average. And — maybe most importantly — teams with starters going on short rest have won 33 times and lost 52 times. Now, maybe these teams are worse, which is why they’re starting guys on short rest in the first place, but they’ve by and large given the ball to their aces, and the aces haven’t delivered like usual. For Don Mattingly, it’s easy to want to write Kershaw’s name in instead of Nolasco’s. But on short rest, Kershaw’s name doesn’t mean what it would ordinarily.

Kershaw prepared for this start over the weekend, just in case he was called upon. He’s been selected as being capable of doing this by his own manager. But that doesn’t make him any different from the other short-rest guys — they all prepared, presumably, and they were all also selected. Sometimes, it’s worked out, but you can’t look at some numbers and ignore the others. Overall, it’s been a poor gamble.

Of course, there are success stories. CC Sabathia was good on short rest in 2009. But, he was bad on short rest in 2008. Curt Schilling was good in 2001. Kevin Brown was good in 1998. Brown was bad in 2004. Roger Clemens was bad in 2000. Tim Hudson‘s been good once and bad once. Tim Lincecum‘s been bad twice. The point has probably been established by now — by no means is it guaranteed that Kershaw will struggle, but he probably won’t be himself at 100%. Kershaw at, say, 90% is still excellent, but he’s less excellent, and this way he’s not available for any potential Game 5.

And there’s the additional factor of what this could mean the rest of the way, should the Dodgers advance. Kershaw’s never thrown this many innings before, and on Twitter, Brandon McCarthy advanced the argument that fatigue could start building due to Kershaw going to work before he’s totally recovered from his last outing. From where I sit, this is just about unprovable, but there’s the possibility that Kershaw could be a little worse over the rest of October because of a start on short rest today. Which, naturally, would be worse news for the Dodgers. But they’re pretty sure he can handle it, otherwise he wouldn’t be asked to handle it.

By starting Kershaw instead of Nolasco on Monday, the Dodgers are increasing their odds of winning Game 4 and moving on. By lining up Greinke to pitch the next one instead of Kershaw, the Dodgers are decreasing their odds of winning a potential Game 5. It’s impossible to say whether the numbers balance out, or come out putting the Dodgers ahead. It’s easy to say that lots of teams have believed in starters on short rest in the playoffs before, and those starters, as a group, haven’t pitched like themselves. This is an unexpected and unusual gamble.

Print This Post

Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.

34 Responses to “Giving a Questionable Start to Baseball’s Best Starter”

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

    I practically had a meltdown during the 2011 WS game 7 when Chris Carpenter started on short rest (I was in attendance). I was well aware of the risks and was constantly afraid he was going to lose it. In the end, he had one of those ‘gut it out’ starts, hardly ace-level pitching but didn’t do too badly and of course we won. Still, I’m not convinced it was the right decision then and I definitely don’t think it’s the right decision for the Dodgers now.

    Vote -1 Vote +1

  2. Helladecimal says:

    So is this Mattingly pulling trigger? Or do you suspect there’s downward pressure from the FO?

    Vote -1 Vote +1

  3. nerdy bill says:

    The best option is to be the 2011 Phillies and not have to be concerned with your third or fourth starter.

    Vote -1 Vote +1

  4. tropicalmug says:

    Couldn’t this be more about the Dodgers lining up the NLCS rotation than anything?

    If Kershaw doesn’t start tonight and they win, then he can do Game 1 on full rest. Great. He can start two games and come out of the bullpen if the series goes long. If Kershaw starts tonight and they win, he can do Game 2 on full rest. Great, you get to keep that advantage in the NLCS. These are the two best cases for each train of thought, that’s why I’m putting them together. You don’t throw Kershaw tonight NOT to win tonight, and the argument for not pitching him tonight is partly based on the understood implication that he wouldn’t be available Game 1.

    If Kershaw doesn’t start tonight and they lose, then they get a fully-rested Kershaw in Game 5 in Atlanta. Where the Braves are 56-25 (.691 winning percentage). Plus, if Kershaw starts Game 5 and they win, he’s not available until Game 3 of the NLCS unless you want to start him on short rest, but you were opposed to that before so why would you do it now. If Kershaw starts tonight and they lose, they still have a pitcher of maybe good Nolasco or Greinke (who really isn’t too far off from a slightly better version of a really good Ricky Nolasco) to throw Game 5, and they were going to be making a start anyway in the other version. Unless you really don’t have confidence in Nolasco at all.

    To top it off, it’s not as though Kershaw starting on full-rest is a guaranteed win. You’re not trading in a sure thing for a not-sure thing, you’re trading in a not-sure thing for a not-sure thing that has a significant advantage in the NLCS.

    Kershaw starts tonight because they don’t think Nolasco would win. They want to give Kershaw two NLCS starts. Well, there’s only one way to do that.

    Vote -1 Vote +1

    • Richie says:

      El wrongo. Unless you’re seriously arguing the Dodgers prefer Kershaw starting 2 games to the Dodgers winning the series. “Well, we won Game 4 but didn’t use Kershaw. Drat!” Dude, you’re really reaching here. Wilt Chamberlainesque so.

      Vote -1 Vote +1

      • tropicalmug says:

        Huh? I don’t say they’ll regret not using him. I’m saying if they use him tonight and they win, that’s way better than not using him tonight and them winning. They think they’re going to win with him tonight, so they’re starting him.

        Vote -1 Vote +1

        • Richie says:

          No, starting Kershaw (fully rested) in Game 1 of the next round and thus getting 3 starts out of him is way better than starting him Game 2 (again on short rest) and only getting 2 starts out of him (with emergency availability in Game 7).

          Vote -1 Vote +1

        • tropicalmug says:

          The only way to get three starts out of him is to start him on short rest. If you wanted to start him on short rest, why not start him tonight?

          Vote -1 Vote +1

  5. Richie says:

    It’s not that unexpected nor that unusual. As shown by the 85-game sample set. Managers typically give in to the pressure this time of year and overmanage, often badly so. Like here.

    Vote -1 Vote +1

    • GoodasGoldy says:

      I agree. This move to me favors the Braves. Mattingly might get away with it but from a Braves perspective they were going to have to face and beat Kershaw to advance anyway. So, getting to face him on less than normal rest for the first time in his career is an advantage whether it pans out or not.

      If the Braves get past Kershaw tonight they will be sky high knowing they are going home for a final game 5 and it won’t be against Kershaw.

      Vote -1 Vote +1

  6. Dave says:

    I think there’s some serious selection bias in your groups for a few reasons:
    1. As you point out, the short rest group is going to be nothing but “ace”-quality pitchers. But the other group is going to be skewed towards better arms as well. “Aces” are going to pitch multiple games every series, while the 4th starters are going to get skipped at every opporunity.
    2. As you mention, something like this is generally a desperation move. That, by definition, implies that usually those starters on short rest are facing teams that have been scoring enough runs to win. Therefore, there’s bias towards better offenses in the short rest group.
    3. I’m guessing the short rest group is a much, much smaller sample than the full-rest group.

    I have no idea how all of this would affect the numbers. Just pointing out an observation.

    Vote -1 Vote +1

    • Richie says:

      1. The short rest group will still be superior in quality to the overall group. Excepting when short resters face each other.

      2. If you’re arguing ‘hot squad’, that’s pretty much been researched away. If instead overall, since they’re all playoff teams, there’s only so much space for quality difference, offense or defense.

      3. The size of the full rest group is pretty much irrelevant. Essentially a matter of the size of the data set (85 games) that you are looking at.

      Vote -1 Vote +1

      • Stevo says:

        Tell me: out of everyone in this very, very small sample, which started on short rest while their team was winning the series?

        1. Inference is in the right tail of pitchers. The distribution is likely to be extremely skewed, so there’s still variation, and so not comparable as you say.

        2. What research? Is it any good? ‘Hot hand’ research has absolutely no consensus, as statistical tests are biased given that they can’t measure latent measures like confidence or concentration, which play a huge role. I’ll give you a reading list if you’d like.

        3. Only clean comparison is against himself, which he has no history of. So most of the points made in this article is purely speculative.

        Vote -1 Vote +1

  7. deadmanonleave says:

    It’s always struck me as panic – If it’s not vital to have ‘x’ amount of days off, why aren’t teams making these changes all the way through the season? If there is a benefit, we’d surely see it before the close season.

    Vote -1 Vote +1

  8. Drew says:

    Kershaw should back out.

    Vote -1 Vote +1

  9. MrKnowNothing says:

    Nolan Ryan approves.

    Vote -1 Vote +1

  10. KidA says:

    In 2008 Sabathia was great on short rest.
    After taking the ball on Sept 16th, he proceeded to take the ball on short rest the 20th, 24th, 28th and then October 2nd in the playoffs.
    The only bad start he had was October 2nd and that was on short rest after throwing 105, 108 and 122 pitches in his last 3 regular season starts. He went 21.2 innings on those short-rest starts, and gave up 2 earned runs. Add in his playoff performance (3.2 IP, 5 ER) and that comes to 25.1 innings, and 7 ER… or a 2.48 ERA.

    I’m of the opinion Sveum went to the well one too many times.

    Vote -1 Vote +1

  11. Z..... says:

    I dont know if I would have done that with Kershaw tonight. He threw 124 pitches in Game 1 and a TON of innings during the year. He hasnt come out on 3 days rest before either. Meanwhile, if they get into the NLCS, they’re gonna need Nolasco at some point and he wont have pitched in over 2 weeks by then. Ricky hasnt been very good on long rest. Look at his numbers this year on 6+ days between starts. I know I mentioned the Braves hitters (namely Brian McCann and Freddie Freeman) owning Nolasco at times, but I would have let Kershaw go out there with his full stuff in Game 5 if needed. Plus, you’re facing Freddy Garcia today. Have more confidence in your offense and a guy you know would have gone out there and got you innings

    Vote -1 Vote +1

    • chuckb says:

      “He threw 124 pitches in Game 1″

      to me, this is key. Mattingly surely had thought out this possibility prior to the series. It begs the question, therefore, why he would leave Kershaw out there to throw 124 pitches with a 5 run lead in game 1 knowing that he was going to bring him back on 3 days’ rest. The Dodgers’ bullpen is pretty good. I realize wanting to bank that first win but you’d have to believe that guys like Bellisario and Rodriguez (who were fresh) are probably better than Kershaw at 110-120 pitches anyway. Plus, add in the fact that he wants to use Kershaw on 3 days’ rest, Mattingly should have pulled Kershaw earlier.

      Fortunately for the Dodgers, Uribe saved Mattingly from having to deal with this.

      Vote -1 Vote +1

  12. Ed Feng says:

    Great numbers on short rest. Do you have the same numbers for strikeout and walk rates?

    Vote -1 Vote +1

  13. Hurtlockertwo says:

    Bye, Bye Braves.

    Vote -1 Vote +1

  14. Stevo says:

    I think this is just bad inference, given the sample size. Jeff has written good pieces, but this one is disappointing, even with his qualifiers.

    A bigger concern is if Kershaw ends up like Hershiser in the ’88 postseason. Lasorda overused him the whole season, and wasn’t the same afterwards. Let’s hope Kershaw doesn’t share the same fate.

    Vote -1 Vote +1

    • bdhudson says:

      In ’89, Hershiser posted his 2nd best WAR year (5.2), which was actually an increase of 1.6 over that ’88 season.

      Vote -1 Vote +1

    • bdhudson says:

      It wasn’t until a year and a half later that he tore his rotator cuff. Easy to point the finger at the ’88 season/postseason, but proving causation and all that.

      Vote -1 Vote +1

      • wally says:

        Certainly the 88′ playoffs wasn’t the only thing doing that. In 1988 Hershiser threw a total of 309.2 innings, then 256 in ’89, oh and in ’87 it was 264.

        While you’ll never prove causation, in that span he racked up more innings in 3 years than most pitchers today will see in 4, while also getting the ball about once every 4.5 games.

        I don’t think there is any question he pitched until his arm fell off, but whether it was the pitching that caused his arm to fall is only a very reasonable assumption.

        Vote -1 Vote +1

  15. Zack says:

    I don’t understand the general consensus within the baseball writing community that Kershaw is baseball’s best starter. To me, Kershaw is bunched up with 5-6 other guys and probably would be behind Harvey in the Cy Young race if Harvey had finished the season and there’s a reasonable argument that Wainwright was better. He pitches in the NL in a great pitchers park, which WAR accounts for and thus places him within that peer group of 5-6 pitchers. Why does it seem that otherwise statistically-inclined writers have anointed him as baseball’s best pitcher this season based on superficial numbers?

    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>

Current ye@r *