A modest proposal for the use of closers

To: All 30 MLB managers
From: Pizza Cutter
Re: Your closer
Hey there, managers!  It’s your old buddy Pizza Cutter from StatSpeak.  You read my stuff, right?  Congratulations on not getting fired this off-season.  (Except for you, Joe Torre.  In your case, congratulations on finally getting fired this off-season.)  I’m sure you’re all looking forward to a new season coming up.  That makes thirty-one of us.  Let’s go (insert your team name here)!
So, I know you guys get this all the time, but I had a great idea that you might want to try in the upcoming season.  No, it’s not something weird like letting Vladimir Guerrero pitch, even though that would be so cool.  Nah, this is actually about your closer.  He seems like a decent chap and he’s your best reliever (in most of your cases).  After all, he racked up all those saves last year.  It’s just that… you’re basically messing up in the way that you’ve been using him.  I mean, not totally.  It’s just that he only ever seems to pitch in save situations in the ninth inning and nothing else.  Sure, he occasionally pitches the ninth of a 7-1 game if he needs an inning of work, and pitches in the 14th inning of a game because someone has to stand out there in the middle of the field and throw the ball toward home plate.  But, usually, you’ve been pretty consistent in these things.  Ninth inning, up by 1, 2, or 3.  If that’s the situation at hand, it’s your closer on the mound, and it’s usually your closer on the mound only in those situations.
I swear I won’t use the word “leverage” here.  Yeah, I’m a stat guy, but since you don’t like the idea of a computer in your dugout, I’ll stick to math that you can do by looking at the scoreboard.  You can subtract, right? 
I suppose that you might have heard some of us stat guys and how we don’t like the save rule or what the position of closer has become.  In fact, bringing a closer in with a three run lead in the ninth inning is actually a waste of a pretty good pitcher.  If a closer is only going to be allowed to throw a certain number of innings per year, why waste those innings in situations where he really isn’t needed?  What if you re-thought your closer as not being someone who slavishly comes in whenever the save rule says he should and instead brought him in when he could actually do the most good?
I did a little math.  It’s no secret that you use your pitchers in reverse order of quality, so I decided to take advantage of that fact.  I looked at all games played between 1990 and 2006.  All of them.  Why 1990?  Because that’s when you all started using the ninth-inning/one-inning closer model that you all fell in love with when Tony LaRussa did it with Dennis Eckersley.  Why 2006?  Because I’m too lazy to add in the 2007 game logs.
For the average team,  in a 162-game season, a team will be going into their pitching half of the ninth with a

  • Three run lead: 11.6 times (and on average, 0.47 games in which the save will be blown, 4.0% fail rate)
  • Two run lead: 14.0 times (1.41 blown saves, 10.0% fail rate)
  • One run lead: 15.2 times (3.69 blown saves, 24.3% fail rate)
  • Save-worthy lead (sum of the above): 40.8 games (5.57 blown saves, 13.6% fail rate)

Now, it’s no secret that you save your second-best pitcher for the eighth inning.  Let’s see how often he faces the same circumstances, just in the eighth inning and how he’s doing.

  • Three run lead: 11.8 times (.66 blown saves, 5.6% fail rate)
  • Two run lead: 14.7 times (1.70 blown saves, 11.6% fail rate)
  • One run lead: 16.6 times (4.34 blown saves, 26.1% fail rate)
  • Save-worthy leads (sum of the above): 43.1 games (6.7 blown saves, 15.5% fail rate)

In general, the difference between an eighth inning pitcher and a ninth inning pitcher is about 1.5%-2% worth of certainty that the lead will be intact at the end of the inning.  (Why the closer makes about $2.5 million more per year than the eighth inning guy, I’m not entirely sure.) 
But anyway, managers, here’s my proposal: bring your closer in to the eighth inning when you are winning by one, and into the ninth inning when you are winning by either one or two.  If you score in either the bottom of the eighth or the top of the ninth and extend the lead, take the closer out.  Let your eighth inning guy handle the 3-run saves in the ninth.  I know some of you just had a heart attack when you realized that this might mean some two-inning saves on the part of your closer and some of you will curiously tell me that closers are not actually capable of going more than one inning.  Perhaps if he knew far enough in advance, Mr. Closer might be persuaded to change his routine a bit in anticipation of going two innings on occasion?  Maybe?
OK, so clearly, we can’t have Mr. Closer pitch 2 innings on three days in a row.  One of the reasons that you love the one-inning closer model is that you can bring him back the next day.  Of course, there are limits to that too.  Managers seem to get a little skittish when the closer is needed four days in a row, so let’s assume that a reliever can only pitch three innings in any four game period.  (Here I’m even disregarding scheduled off-days, which don’t count in the standings, but still count as a day off).  How often would the closer not be available?  About three games per year (2.93, to be exact).  Of those games in which the closer was not available, the number of times his presence was needed?  About 14.7% of those games.  On average, not even half a game per year.
Now, why bother switching things up like this?  First off, you can’t give me the excuse about “defined roles.”  Plus, since you’d make the changes at the beginning of the inning, just have the closer (or the eighth inning guy) warm up in the half-inning before.  Now, consider that basically you’re asking your closer to trade his ninth innings with three-run leads to the eighth inning guy in exchange for the setup guy’s one run leads in the eighth inning.  The setup guy is 1.6% worse than the closer at protecting 3 run leads, and he’ll have 11.6 more to protect, so he’ll net 0.1856 more blown saves.  However, the closer is 1.8% better at protecting 1-run leads, and he’ll have 16.6 more to protect, which will net you 0.2988 fewer blown saves, for a net gain of 0.1132 fewer blown saves.
Wait a minute.  All that for a paltry shot at maybe one fewer blown save?  Well, yeah.
Let me guess.  You’re worried about the one day when you don’t have the closer available, the setup guy inevitably blows it, and you’re whipped by the media for “riding your closer too hard” and adopting such an “against the book” strategy?  After all, your job performance is based on the “what have you done for me lately” principle.  And while making the team a little better is always nice, “the book” is easier to follow because no one ever blamed the manager for following the book. 
Forget we talked.  This message will self-destruct.

About these ads

One Response to A modest proposal for the use of closers

  1. David Hannes says:

    Another variation might be to have two closers–one righty and one lefty–and choose the one where it makes more sense, e.g. if the opponent has two righties and a lefty due up in the 9th, but two lefties and a righty due up in the ninth, then go with the righty for the 8th inning and the lefty in the ninth. Yes, I know pinch hitters ruin the probabilities…but they’ll do that with the closer and set-up men now…and by forcing the opponent to go to their bench earlier means fewer choices if they should rally in the ninth…as well as the chance for worse defensive players in the field for an inning.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

%d bloggers like this: