The blown save

There have been 1,672 save opportunities this season, of which 1,150 have been converted. Simple math reveals, then, that there have been 522 blown saves in 2012. Of those 522 blown saves, 321 (61.5 percent) have occurred before the ninth inning.

Now, to put any of that in the proper context, it should be examined what a blown save is. A blown save is charged to a pitcher who enters a game in a situation which permits him to earn a save, but who instead allows the tying run to score. Such a situation includes a pitcher entering the game with a lead of no more than three runs and pitching for at least one inning, entering the game with the potential tying run either on base, at bat, or on deck, or pitching for at least three innings to close out a game.

To date, Marc Rzepczynski has five blown saves this season; he has pitched with a lead in the ninth inning or later five times, and only once in a save situation. Rex Brothers has five blown saves this year; he has pitched with a lead in the ninth or later five times and in only one save situation. Despite their lack of appearances with a ninth inning lead, both pitchers find themselves among the leaders for blown saves on the season, along with one-time closers Heath Bell, John Axford, and Santiago Casilla.

It occurs to me, then, that there may be a flaw in the blown save statistic. Essentially, dissimilar events are being compared within the same category. The solution I submit (for the approval of the midnight society) is to re-categorize blown saves occurring before the ninth inning.

To be clear, this isn’t meant to be any sort of referendum on saves or who should be eligible for saves. In fact, I have no issue with the save statistic. Its counterpart, though, seems to cast too wide a net. As it stands, blown saves don’t discriminate between closers and middle relievers. I think it is of little consequence to establish closers as one inning, the ninth, specialists. It seems odd to me to be measuring their failures with the same tool as other relievers. Seeking to redefine the blown save is simply a reaction to the modern use of closers.

I realize that I have discussed closers a great deal so far, when, in fact, my motivation for investigating this was to separate middle relievers from closers. It just seems to me, though, that the best means of properly explaining how and why to distinguish one blown save from another is by looking at the closer role first.

It stands to reason that losing a lead in the ninth inning or later is very different from losing a lead in the sixth to eighth innings. Quite frankly, losing a lead before the ninth inning still allows a team an opportunity, or multiple opportunities, to bat and, in theory, regain their lead. Losing a lead in the ninth inning or later does not guarantee such an opportunity, unless the home team has lost their lead. The consequence of a home team losing a ninth inning lead, though, is an extension of a game that should have otherwise ended. This is where I think the definition of the blown save should be directed. I propose that a blown save should be connected directly to a loss or the otherwise unnecessary prolonging of a game.

So, what to do with other lead losses that fit the current criteria of the blown save? I propose an added category to be called “blown holds.” Blown holds would cover blown saves taking place before the ninth inning. If completely redefining the blown save is too radical, then consider this an addendum to the category, an easier means of differentiating the timing of a lost lead. At the very least, such a statistical category would provide a more proper counterbalance to the mostly overlooked hold statistic.

To return to my original player examples, four of Rzepczynski’s five blown saves would be categorized as blown holds; the same would be true of Brothers. Both players pitch for teams with, relatively, stable closers. In fact, only one player, Jason Motte, has recorded a save for St. Louis. Both the Cardinals and Rockies have double digit blown save totals, though. St. Louis has blown 22 of 58 save opportunities; 14 of those blown saves have taken place before the ninth inning. Colorado has blown 30 of 62 save opportunities; 18 of those would now be considered blown holds.

The current blown saves leaderboard is littered with middle relievers who have had few or no legitimate opportunities to close out a game in a save situation. A leaderboard more befitting of middle relievers might be a hold percentage leaderboard. Hold percentage would, quite simply, be calculated by dividing holds by hold opportunities, which would be the sum of holds and blown holds. Particularly confounding must be Heath Bell’s success before the ninth inning, especially when contrasted with his ninth inning struggles.

Leaders*

Player           Team          Holds Blwn Hlds  Hold Opps  Hold %
Randy Choate     Mia/LAD         20         0        20    100%
Grant Balfour    Athletics       15         0        15    100%
Juan Cruz        Pirates         14         0        14    100%
Ryan Mattheus    Nationals       13         0        13    100%
Andrew Miller    Red Sox         13         0        13    100%
Jerry Blevins    Athletics       12         0        12    100%
Joe Thatcher     Padres          12         0        12    100%
Heath Bell       Marlins         12         0        12    100%
Santiago Casilla Giants          10         0        10    100%
Ryan Webb        Marlins         10         0        10    100%

Trailers*

Player                Team            Holds  Blwn Hlds  Hold Opps  Hold %
Kameron Loe           Brewers            6          5        11     55%
Andrew Cashner        Padres             6          4        10     60%
Matt Albers           Bos/AZ             8          5        13     62%
Jesse Crain           White Sox          9          4        13     69%
Tim Collins           Royals             9          4        13     69%
Aaron Crow            Royals            18          6        24     75%
Fernando Rodriguez    Astros            12          4        16     75%
Chad Qualls           Phi/NYY/Pit       13          4        17     77%
Alexi Ogando          Rangers           11          3        14     79%
Rex Brothers          Rockies           15          4        19     79%

*minimum 10 hold opportunities

It’s my hope that the introduction of the blown hold can help add some weight to the blown save statistic. While the blown hold can be seen more as hiccups in a game, perhaps the blown save can be viewed as truly detrimental to a team’s season. I believe differentiation can only be of help in this instance in the statistical narrative. Either way, this is a simple way of shaping the discussion about relievers who successfully, or unsuccessfully, hold leads.


11 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
hopbitters
11 years ago

Just to clarify, the relievers in question were in a potentially save-eligible situation and lost the lead before the 9th?

bucdaddy
11 years ago

Did you know you can get a hold and a loss in the same game? Saw that one in a box score this year and did a double take.

Even before that, I thought the hold was kind of stupid. So I don’t think much of blown holds, sorry.

David P Stokes
11 years ago

The real problem here is that for non-closers, save opportunities usually aren’t save opportunities.  Your starter goes 7, you’re up by 1, and you bring in the setup man to pitch the 8th.  If he loses the lead, he gets charged with a blown save.  But if he doesn’t blow it, he almost never gets a chance to actually earn a save—he just gets a hold and passes the save opportunity on to the closer.

I’m not convinced that we need a new stat to quantify this, though—an awareness that for non-closers their ratios of saves plus holds to blown saves is the rough equilalent of the ratio of saves to blown saves for closers should suffice.

Paul G.
11 years ago

I do agree that the Blown Save stat leaves something to be desired.  However, there are a couple of issues that I think would need to be addressed with the Blown Hold.

First, you need to define what a Hold is.  Last I checked there is no official definition and there are multiple unofficial ones.  The differences are subtle but can be significant.  Oddly, it may not matter for the Blown Hold as most of the differences involve whether the pitcher did enough to earn a Hold.  Failure is a bit easier to define.

Second, while the 9th inning rule generally works for modern day closer use, it does not for closers of days past.  It used to be common for the relief ace to enter in the 7th or 8th inning and finish the game.  If Bruce Sutter blows the lead in the 7th that’s a Blown Hold under this rule set, but Bruce was most certainly expected to finish this thing so it is a Blown Save in spirit.  Even today the 4-out Save is not extinct.  You may be able to adjust this slightly so that if the pitcher finishes the game or at least completes the 9th inning on the way to extra frames then he gets a Blown Save rather than a Blown Hold.  It wouldn’t help in situations where the closer gets mauled in the 7th and gets lifted though.

David P Stokes
11 years ago

The definition I generally see for a hold is that the reliever enters the game in a save situation and passes the save situation on to another reliever while recording at least one out.

Mike Schryver
11 years ago

It would make more sense to work toward counting Saves and Holds together than to create yet another statistic.  That would address the problem you’re describing.

Paul G.
11 years ago

Looking at the obscure Hold rules, #1 or #2 shouldn’t impact the Blown Hold at all, #3 and #4 would need to be addressed, and #5 probably doesn’t matter though I suppose it could under some scenario.  All 5 rules impact the Hold itself which will be important for the success ratios.

Now if you are simply using existing Hold/Blown Save data calculated by some source, then the BS to BH conversion should be fine.  Simply keep in mind that the results from your source may not match another.  If you decide to calculate this from scratch then things get more complicated.

Paul G.
11 years ago

@David
A couple of years ago I tried to calculate Holds from Retrosheet data and quickly discovered that various sources (baseball-reference, Elias, STATS, etc.) had, in some cases, different totals for the same pitcher/season.  For that matter, STATS in 2001 had different totals than STATS in 2009.  This made vetting my Retrosheet project a bit more difficult as you might expect.

While the general definition of a Hold is as you say, there are some unusual scenarios that are treated differently by different sources and generally are undocumented.  Here’s the list of differences that I found:

1.  Does the pitcher need to get an out?  (Yes, there was one source where getting an out was not required.  Why they thought this a good idea is beyond me.)
2.  Can a pitcher get a Win and a Hold in the same game?  (This would apply if the starter was lifted with the lead but before completing 5 innings, allowing the official scorer the discretion of awarding the Win.)
3.  Related, can a pitcher be awarded a Hold if he enters in the 5th inning or earlier? 
4.  For that matter, can the first reliever (the guy after the starter) who enters before the 5th inning in a Save situation get a Hold?  Technically, he is the tentative pitcher of record at that point.  If he finished the game and held the lead he would get the Win.  This gets all mixed up with #2 and #3.
5.  Can a pitcher get a Hold for pitching 3 innings of effective relief?

For all 5 of these examples, there was at least one definition that said “yes” and at least one that said “no.”  So you do need to be specific on this.  The differences are typically small and most pitchers will have the same total, but there seems to be a few guys every season where it makes a significant difference.

adam hayes
11 years ago

first of all, thanks to everyone for reading.

i just wanted to take a moment to clarify and reiterate some things. all my blown holds data came directly from mlb.com box scores. the blown save and hold totals came from fangraphs. so, the data is consistent, as far as i’m concerned.

what i really wanted to address was the idea that holds and saves being calculated together was a better solution. i respectfully disagree with that concept. my point in proposing the blown hold, as i wrote in the article, was to help differentiate between reliever-types. my issue with hearing about the blown save has always been the assumption that it leads to losses. that 60+% take place before the ninth inning is enough evidence to me that that isn’t necessarily so. my hope is to see a more direct consequence (loss or game extension) attached to the blown save, especially since the save is directly connected to a consequence (the last out of a win within certain parameters). separating blown saves as we know them now into blown saves (ninth and later) and blown holds (pre-ninth), to me, is a telling, short-hand means of reliever usage and successes/failures. back to the original point of this paragraph, combining holds and saves for a percentage purpose only continues to muddy the waters of reliever usage. i understand the purpose of such a method, particularly if the aim is to completely devalue/eliminate the save statistic. but that leads me to ask if the issue, then, is with the save (a statistical record of games finished with a close lead, mostly) or how games are actually played and managed to accommodate the statistic (what i see to be the real problem, which players (closers) and their agents are absolutely in favor of maintaining given the monetary incentive currently on the statistic)?

finally, if my idea hasn’t been completely disregarded, i’m more than willing to put my data up on google docs if anyone has any interest in viewing it.

thanks again for reading.

too much nonsense
11 years ago

“Of those 522 blown saves, 321 (61.5 percent) have occurred before the ninth inning.”

So there is little point to the Save in the first place.  I didn’t bother to read the rest, and I wonder if you even know what the least likely inning is for a lead change to occur…and why.

Fake Chanel
8 years ago

What we have here is the standard 2.55 Reissue Flap with Mademoiselle chain, rendered in perfect patina’d alligator. It’s from the Pre-Spring 2010 collection, it’s utterly gorgeous, and it’s out of my price range.