Saving Games

Jim Caple has an excellent column up now on  Reading it reminded me of something I’d been meaning to write for a while.  I’ve mentioned this before in the context of bashing Joe Torre, and I thought I’d expand on it a little.

Caple’s angle is basically that the “save” is the most overrated statisticin baseball, and that the “closer” is the most overrated position.

Sabremetricians among you may be familiar with Bill James’s “Bullpen Ace” model.  What I find interesting is that people often confuse this model with “Closer by Committee” which is, in fact, wrong.  According to James himself, he doesn’t like that model of bullpen management.  From his 1984 Baseball Abstract:

I’m a little skeptical about group bullpens in principle . . . if you don’t have a bullpen ace, things can get awfully confused sometimes; one pitcher gets into a slump and then another and another, and you don’t really know who it is that is supposed to get you out of this. I like definition in a pitching staff; I like a staff with four starters, a relief ace, a middle-inning man, a spot starter/long man, a lefthanded spot reliever, a mop-up man. I like that . . . it is easier to find five guys who can pitch than it is nine or ten. When you have a group bullpen, you’re going to have your #8 pitcher out there on the mound with the game on the line 30 or 40 times a year. I don’t like that. It also means that you have to find 8 or 9 effective pitchers, and I don’t like that.

What I think is most interesting is that he refers to a “relief ace” but not a closer, and that the situation that he thinks is the problem (ie: the one thing to avoid regardless of the model) is when you have your #8 pitcher out there with the game on the line.  Ironically, while many people would ignore James’s model in favor of the current in vogue “closer” model, it’s the closer model that yields this outcome more often than not.

Take a hypothetical situation, though I’m sure similar situations have occurred countless times in reality.  The Yankees are playing the Red Sox in Fenway.  The Yankees have a one-run lead in the late innings, and Pedroia has just walked, with Manny (thank God not anymore, so pretend this is last season), Ortiz, JD Drew due up.  Let’s assume your starter is already out of the game.  If you’re the Yankees, which pitcher do you want out there?  If your answer is anything other than Mo Rivera, you’re an idiot.  And if your answer starts with “what inning is it” you are not absolved from being called an idiot.  Does it matter if it’s the seventh, eighth or ninth?  It shouldn’t.  This is the critical time.  These are the guys that are going to beat you.  Don’t you want your best pitcher facing them?  Let’s assume it’s the 8th.  Do you really want to pitch Kyle Farnsworth or Edwar Ramirez against these guys, so you can bring in Rivera to start the 9th with nobody on and to face Julio Lugo, Jason Varitek and whoever is batting 9th for the Sox?  Of course not!  You pitch your best pitcher in the most important time of the game.

But that’s not what accepted baseball convention is.  Accepted baseball convention would say that you find someone to get through the 7th and 8th regardless of situation, and bring in your “closer” for the 9th.  Even if it means they never come in with men on base, and often face the bottom of the lineup for three easy out, at a point in the game when the outcome is almost a foregone conclusion anyway (see Caple’s article for the breakdown of stats, that in almost every decade from the 1900’s to today, teams’ winning percentage with a lead in the ninth has been remarkably consistent, from before the invention of the closer and after).  Francisco Rodriguez has like 7,000 saves already this season, but hasn’t come into a game before the 9th and has never come in with men on base.  Does Mike Scoscia mean to tell me that his team has never been in a situation when the game was on the line before that?

I don’t mean to pick on Scoscia, because it’s accepted convention already.  But someone has to have the cojones to buck the trend.  But what’s also part of the problem is the statistic.  Closers and their agents want to make sure they rack of the “saves” and the only way to do that is to pitch the 9th.

Why can’t baseball redefine the “save” to give the official scorer some discretion (which he already has when it comes to deciding wins, errors, etc.) to decide when the most important relief moment of the game was, and award the save to the pitcher who pitched out of it, even if it wasn’t the 9th.  I guarantee that would change the players/agents/managers style very quickly.


18 responses to “Saving Games

  1. Note that that’s what the good teams essentially already do. The best recent example was last year’s Indians (I’m most familiar with them anyway 🙂 ). Their best two relievers were Perez and Betancourt, who would come in in every sticky situation possible and get out of it. They were the “aces”. Borowski was a guy who came in to pitch one inning, usually with a 3-run lead, and would give up a run while finishing the game. He wasn’t generally pitching in any difficult situations. Even this year, with the extensive troubles the entire Indians ‘pen has had, they still use them wisely – Perez has rebounded after a rough start and is now used in closer games again. Koboyashi is the so-called “closer”, when it’s close, but he’ll come in in the 8th (like he did last week) or they might leave whoever is pitching in to get the save (like Perez did a couple nights ago). Betancourt has struggled, so he’s not in as many tight games, but he’s still generally used like that.

    I think the Cardinals also do similar things.

    The Save isn’t *completely* overrated, as presumably 2/3 of save opportunities come with a 1-2 run lead, where you can’t afford to have a bad pitcher allow a couple baserunners or a HR. It’s why the Indians eventually tired of Borowski, and why closers are generally very good pitchers. However, the save as a stat is a horrible measure of how good someone is; it just means that using good pitchers in those situations isn’t completely farfetched.

  2. Ezzie,

    I don’t think “good” teams do it, and certainly not by design. Look at the Yankees with Mo, the Red sox with Papelbon, the Angels with K-Rod, the Twins with Joe Nathan, even Wagner with the Mets. All of them would be considered the best reliever on their respective teams, and they are penciled into the 9th inning without so much as a thought on the situation.

    Even your example of the Indians from last year, likely wasn’t by design. They spent a lot of money to sign Borowski to be their closer, with the anticipation that he’d be their best reliever. The emergence of the other guys gave them a good bullpen, but that wasn’t the plan.

  3. noyam – i’ve been saying this for a few years. couldnt agree more.

  4. Fangraphs carries a stat called pLI which stands for the player’s leverage index (

    It does exactly what you are talking about and is usually a good predictor of who is “next-in-line” when a closer goes down and who is only getting mop up duty.

    I don’t know about other teams like I know the Jays but obviously the two guys with the highest pLI are Ryan and Downs. Jason Frasor, who only gets work if every other is tired or the game is out of reach, has the lowest pLI among relievers.

    Using this stat (and assuming that you accept the premise of how it’s calculated), Borowski was using in the highest leverage situations for the Indians last year.

  5. so many grammatical errors there. sorry.

  6. This is what ticked off Goose Gossage for all these years. Poor guy used to enter the game in the 6th inning and pitch for 3 innings at a time…on back to back nights.

    “Athletes” these days can only pitch 1 inning at a time, and most have their own (non-Joba) rules where they get a day off in-between. Perhaps if clubs stopped babying these pitchers, you’ll find pitchers pitching 15-20 years becoming the norm as opposed to the exception.

    Aside from the ridiculous innings limit we see these days, don’t you think pitchers should be on a yearly pitch count, and not the 150 innings or so rookies are limited to? Makes more sense to me. If Joba throws 5 pitches each inning, should he still be limited to 120-150 innings? Or what if he throws 30 pitches/inning. Should we still allow him to reach that inning limit?

  7. Mike,

    It’s definitely an intersting stat, though I would be interested in the methodology for coming up with the leverage tables, because I’m not sure how they get those numbers. For instance, as a general rule, they increase the leverage with 2 outs than with 0. While this my be true for the hitters, wouldn’t the opposite be true for pitchers? Shouldn’t there be a “zero sum” quality to the leverage between batter and pitcher?

  8. are you basing that on these tables?

    to be honest, what youre saying makes sense. and i dont know – i usually dont look at these sorts of formulas (i.e. pLI, EqA, VORP, ERA+, OPS+, etc…), instead assuming that the people that created them are much better at math/stats than i could ever hope to be. and since this particular stat matches up very well with what i “know” (i.e. downs in every tight situation, frasor in garbage time), i just assumed that it was valid.

    could it be, though, that leverage increases as the game winds closer to an end because the pitcher’s offense has a diminishing chance to score runs (i.e. as each out moves closer to the 54 or 51 outs required to end a game, leverage increases)?

  9. Yeah, that’s the table I meant.

    I also think that the leverage table doesn’t necessarily tell the whole story, becuase it doesn’t factor in the batter.

    There has to be some recognition that facing David Ortiz in the 8th is higher whatever-you-want-to-call-it (leverage, pressure, difficulty, importance, whatever) than Julio Lugo in the 9th.

    So yes, as you approach the end of the game, every out means more. But again, pitching in the bottom of the 9th, bases loaded, up by 1. Wouldn’t you rather have 2 outs than 1 and 1 rather than 0? That has to be some sort of baseline assumption, right?

  10. Interesting post. The most common rationale for the current “closer” trend is that closers are only comfortable coming into the game in a closing capacity. Many managers have been vilified for bringing in their closer in a tied game or when their team was down a run b/c the closer was like a deer in headlights in that situation. I wonder how the stats look in situations where closer was brought into a game in a non-closer situation compared to that closer’s overall stats. Could it be that someone like Mariano would be a closer to average pitcher if he was regularly brought into games down one run in the seventh? Could it be that some middle relievers who regularly pitch 2+ innings would have far better stats if they were regularly brought in in the ninth to save games where their team was up three runs?

  11. Please don’t get mad at me for cross posting or whatever it’s called. Today’s Mets game is an example of Santana getting bailed out by his bullpen. (Of course, the Mets have only scored three runs so far, but he had a 3-1 lead and nearly let it get away, but the bullpen shut the door.)

  12. Spoke too soon…

  13. Adam,

    Even before the bullpen blew up on him (again) I thought that Santana’s start today was an example of why win-loss records are not a significant stat.

    Dude gave up 1 run in seven innings of work. Was sent out there to pitch the 8th, probably should have been yanked as soon as he gave up the first hit (but that’s a manager issue) and then the ‘pen allowed a run to score on his behalf.

    When a pitcher goes 7 innings and gives up one run, that should be considered the kind of start that generates a “win”. It’s not his fault that in 7 innings of work all the Mets could muster was 3 runs of support.

  14. I don’t agree. I think that when you *only* pitch seven innings, you give your team a chance to win but also give your team two innings to lose. He would have deserved the win if he had gone nine innings and only given up one run. His lack of stamina is a key reason why he didn’t get the win today.

  15. I shouldn’t say “I don’t agree.” I do partially agree.

  16. I think you aren’t being realistic as to how deep into a game a starter typically goes these days. Overwhelmingly, I would suspect, starters that give up 1 run over 7 full innings of work, earn a win during that start.

  17. “But again, pitching in the bottom of the 9th, bases loaded, up by 1. Wouldn’t you rather have 2 outs than 1 and 1 rather than 0?”

    Yes. What you’re saying makes sense. But being in the same situation with a 3 run lead and no outs is a higher leverage situation with 0 outs than with 2 outs, according to the table.

    Perhaps it’s based on the likelihood that teams score runs in those situations, using the last 100 years as evidence as to how high-pressure the situation is. I don’t know.

  18. I agree that “these days” pitchers don’t go deep into games – but that’s also why I think they are less deserving of a win. A pitcher who pitches seven innings and gives up one run deserves the win XX percent of the time. There is some (probably predictable) ratio for how often a bullpen will blow a lead in those games.

Leave a Reply

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

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

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s