Skip to content

How Geren Has Used Bailey and Weighted Saves

March 19, 2011

A commenter, JK had the following comment:

“I imagine the 4 and 5 out saves are the highest leverage situations as the closer would not have been brought in unless the sh*t had already hit the fan. It might be interesting to go back and look at saves from 2010 and see how many were of n # of outs. Maybe there are some “under-rated” closers out there who have lower total Saves but a higher percentage of 4 out saves.”

I largely agreed with him, thinking it must be true. So, I undertook a very small study to see how Geren used Andrew Bailey and the other two relievers who filled in in his absence Michael Wuertz and Craig Breslow. What was interesting was that I saw a lot about how Geren used his closers and found some surprising data. The statistic I used to try and measure the “underrated” aspect of the closers is a weighted save. I don’t know if this is a real statistic anywhere (I haven’t found it but I also don’t feel it is possible I am the first to stumble upon the idea) but here is how I calculate it. Simply put you first take any outing that resulted in a save and then add up the gLI (game leverage index when the pitcher enters the game). For extras I am also giving you an average save to better understand what the game leverage was like when that pitcher entered the ballgame in his save situation. For basis of explanation I have included the three A’s pitchers with five or more saves along with Brandon League of the Seattle Mariners.

Pitcher WSv AvSv
A. Bailey 44.52 1.78
B. League 18.24 3.04
M. Wuertz 15.11 2.52
C. Breslow 13.78 2.76

As expected Andrew Bailey having more saves (25) has a higher weighted save number, it is essentially a counting stat so that is expected. Then we have Brandon League and Michael Wuertz both of whom had six saves, yet it is clear using a weighted save that Brandon League had much tighter situations to contend with than did Wuertz when he took the hill in position to get a save. Craig Breslow had five saves. But what we also see that is interesting is that Bailey had the easiest saves (lots of those three out variety saves mentioned above by JK), Wuertz’ were more difficult, and Breslow’s were even more difficult than what Wuertz faced. Now one flaw in my thinking here may be that I have excluded blown saves, but I did so because Breslow and Wuertz had blown saves when they clearly were in a technical save situation but truly were being used in a hold fashion, not sure how to handle that so I omitted them, but presumably the leverage index could change with these blown saves situations (particularly for Andrew Bailey). Take a game like May 11th when the A’s at the Rangers had Bailey come in to lock up a 5-4 lead in the ninth. He entered the game with a 3.54 gLI, but a Justin Smoak walk which eventually was sacrificed and moved over on a ground out came to score on an Elvis Andrus hit tying the game. So that 3.54 is clearly a save situation and a high leverage one, but it isn’t counted, so there’s that caveat to these findings.

But regardless it still showed some interesting facts, those eighth inning saves the sort Rollie Fingers and Rich Gossage made legendary (though many of theirs began in the seventh even) for Bailey were often sort of easy. Yes he had a very tough situation to contend with when in Boston on June 3rd he came in (gLI 4.22) to face Kevin Youkilis in the eighth with men on second and first and no one out sporting a two-run A’s lead. But then he had outings like his one that came right before this one, where Geren brought him in with a three-run lead with a man on first to face Carlos Guillen (gLI 0.75).

Other times Geren seemed to fall into the Mike Scioscia get-your-closer-a-save-no-matter-what mentality like on July 18th in Kansas City. Oakland had a 9-5 lead with two-outs in the bottom of the ninth and Jerry Blevins had just allowed a run-scoring single to David DeJesus, narrowing the A’s lead to 9-6. So now with DeJesus on first, technically the tying run was in the on-deck circle which equals save situation. Geren marched to the mound, removed the southpaw Blevins and Bailey (gLI 0.52) came in and got Billy Butler to fly out to right earning the save.

When I commented to JK’s comment I said how I was actually surprised that the closer had such high leverage indexes typically highest on the club. But it seems like with Bailey at least a lot of this is him himself ratcheting up the leverage. I suppose this is natural because every pitcher will give up hits, walks, etc and for a closer that instantly creates a problem where for a starter in the second inning it is much less of a big deal, but Bailey came into his successful save situations with a gLI of 1.78 and oftentimes he was responsible for pushing it upwards (admittedly not significantly) as his pLI for the year was 1.82.

I think the “weighted save” is a potentially useful mechanism for seeing the value of saves, because looking above we see that Brandon League‘s saves were distinctly more difficult than those of Wuertz though in the end they earned the same six. Several ideas I am bouncing around are perhaps subtracting the gLI of blown saves from the total, to penalize pitchers who blow games in very tense situations but this seems to me that it’d be unfairly harsh on middle relievers who earn a few saves, among the top seven in 2010 for blown saves, five of the seven had more holds than saves indicating to me they were middle relievers who would be harshly penalized by this sort of statistic, specifically guys like Daniel Bard (seven blown saves to three saves) or Matt Guerrier (six blown saves to one save). Any ideas are welcome, but I think this is a reasonable starting point.

2 Comments leave one →
  1. March 20, 2011 2:59 pm

    Thanks for that analysis. Very interesting indeed and now that you brought it up from the question of how Geren uses his closers it might be fun to look at a few managers that have had multiple closers over their managing careers and if they treated each closer the same (an over-arching philosophy) or used the kid gloves with some closers and not others.

    I do like the idea of counting blown saves somehow, but I agree that would be trickier than first blush. I do think it is more useful look at leverage when entering the game rather then peak leverage which may be the closers own damn fault. One more crazy idea that just popped into my head. What about some sort of Delta leverage. Maybe as simple as (peak leverage)-(leverage when entering the game) to look at which closers make you run for your anxiety medication: Brian Wilson we are looking at you.

    • March 21, 2011 10:40 am

      JK – I really appreciate your comments, as I find bullpen usage strategy a crazy albeit narrow field of interest for me, so I like your idea about seeing how managers utilize different relievers and seeing if we can tweak it out via stats. I think Art Howe is a good candidate for such a project as his Houston time was largely split between Dave Smith and Doug Jones, and then in Oakland he had Billy Taylor, Jason Isringhausen and Billy Koch as closers then in New York had both Armando Benitez and Braden Looper, a good range of pitchers.

      I think the method for making it fairest would be the following. First defining a closer as someone who Sv>H+GS. That way you whittle out any starting pitcher who sometime late in the season came in and managed to get a save or something, you whittle out guys who are clearly middle relievers too. There may be an outlier here or there for some September call up but if you put a cap on it with a minimum of five saves or something. Next, I think you then do a formula that is the (total)gLI(in saves)-(total)gLI(in blown saves). That way you have already teased out the closers and can now penalize them for screwing up high leverage situations. I think that way is fair, because otherwise you are forced into guessing a manager’s intention which is impossible.

      I like that idea too of the shaky closers, the aforementioned Armando Benitez I am pretty sure would go into that category too. However I think instead of peak leverage here I’d look at average leverage for the game (pLI) because if Wilson loads the bases let’s say with no one out that’s high leverage, but it is still higher should he strike out that next hitter, and significantly higher than had he just left the bases empty.

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 )

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

%d bloggers like this: