Why K-Rod and Putz Might Not Matter

Ask anyone why the Mets finished in second place last year and the immediate answer is “the bullpen stunk”. People are quick to point out the 29 blown saves as evidence supporting that claim. Also buying into that theory was the Mets’ front office, who sought to band-aid the problem by acquiring the AL West’s two best closers. Problem solved, right?

Not so fast. Before we assume that J.J. Putz and Francisco Rodriguez slamming the door on innings 8 and 9 are the “final ingredient” for the Mets’ entry into the postseason, let’s continue to follow the data.

Blown Saves: Putz and K-Rod

Question: who blew more saves last year, J.J. Putz or Billy Wagner?

Answer: You may be surprised to find out that Putz blew 8 games, to Wagner’s 7. But Putz was injured in 2008, so we’re willing to give him a pass. Right?

Question: who blew more saves last year, Francisco Rodriguez or Aaron Heilman?
Answer: K-Rod, who blew 7 to Heilman’s 5.
Granted, K-Rod converted 89.8% of his save opportunities, finishing with 62.

But still, 7 blown saves is 7 blown saves. Add Putz’s 8, and the Mets acquired 15 blown saves this offseason — more than half of the 29 they blew in 2008.

Fans will find out quickly that despite their skills, Putz and K-Rod are not “automatic”. In fact, of K-Rod’s 68 innings pitched last year, he went one-two-three only 22 times (FYI, the Royals’ Joakim Soria led all of MLB with 36 “clean” innings). Also of note: K-Rod never pitched more than one full inning in 2008.

Breaking Down the Mets’ 29 Blown Saves

A few numbers to consider regarding the 29 blown saves that supposedly ruined the Mets season:

9: the number of games that were WON by the Mets, in games they blew a save
13: the number of blown saves that came after Billy Wagner went on the DL
11: the number of blown saves that occurred BEFORE THE 8th INNING

That last number is most intriguing. Many people don’t realize that a pitcher can be assigned a blown save as early as the 6th inning. The big deal about getting Putz and K-Rod is that the Mets can now “shorten the game” to 7 innings. However, the Mets will still have to find a way to bridge the gap in the 6th and 7th, a time when more than one-third of their blown saves occurred.

Subtract those 11 “early blown saves” from the 29, and you’re down to 18 blown saves. Subtract the 9 games that were won, and you’re down to 9 blown saves that occurred in the 8th or 9th inning, that resulted in a loss.

Suddenly, the Mets’ bullpen doesn’t look so awful, does it? Now, consider again that Putz and Rodriguez combined for 15 blown saves last year, and ALL of their blown saves occurred in either the 8th or 9th frames, and you tell me whether the bullpen is definitely improved over last year.

Joe Janish began MetsToday in 2005 to provide the unique perspective of a high-level player and coach -- he earned NCAA D-1 All-American honors as a catcher and coached several players who went on to play pro ball. As a result his posts often include mechanical evaluations, scout-like analysis, and opinions that go beyond the numbers. Follow Joe's baseball tips on Twitter at @onbaseball and at the On Baseball Google Plus page.
  1. upson March 25, 2009 at 6:16 pm
    Joe, I’m glad that you’re bringing up this issue, because I’m getting tired of how the “29 blown saves” argument is being used. You correctly point out the two most important facts the fans and annoucers tend to “forget”: (1) Mets won a lot – exactly 9 – of those games; (2) Many blown saves occured early in the game. On the other hand, I believe that you got carried away in your reasoning, and used a few invalid arguments as well:

    (1) you cannot subract all wins and all early blown saves from 29. Keep in mind that some of those wins came after early BS, hence you’re subtracting some games twice. Hence, it is NOT true that there were only 9 BS in the 8th/9th inning resulting in a loss. There were in fact 12 such games. Here’s the breakdown according to my records:

    8th inning: 7 (2W – 5L):
    9th inning: 10 (3-7):

    The losses came in games #62, #84, #106, #136, #146 (8th inning BS), and #46, #65, #100, #110, #118, #133, #147 (9th inning BS).

    (2) You’re comparing 9 games (correctly should have been 12) lost in 8/9th inning with 15 BS by Putz and K-Rod. This is not correct, since some of those 15 BS by Putz/K-rod resulted in wins by Seattle/Anaheim. Hence, you should have compared Mets’ 17 BS (in 8/9) to 15 BS by Putz or K-rod. [Or, alternatively, if you want to consider actual losses only, then 12 Mets’ losses to whatever number of games were actually lost by Seattle/Anaheim.]

    (3) The absolute numbers are not that important. Putz and K-rod blew 15 saves out of 92 (69+23) opportunities. And this includes subpar efficiency by Putz in 2008. The Mets’ 17 BS most probably were generated in considerably less number of opportunities. [Although the exact number is impossible to calculcate given that a pure set-up man entering the 8th inning does not have a chance to achieve a save, but will be charged by BS if he fails to get a hold.]

    (4) BS statistics do not say the whole story… For instance, how many games tied in 8-th and 9-th were lost by Mets’ relievers? How many such games were lost by Putz and K-rod? I actually did a quick analysis and found out that Putz “saved” 7 out of 9 ties and K-rod “saved” 4 out of 5 ties. I’m sure Heilman alone had at least 5 “blown ties” late in the games.

    Put together, I agree that constant higlighting of “29 blown saves” is ridiculous. On the other hand, having K-rod and Putz for 8-th and 9-th inning last season would definitely be enough to get the Mets into the postseason.

  2. joe March 25, 2009 at 10:06 pm
    Upson, some good points you raise, but I have some retorts:

    (1) – I like the way my math works better. It’s more dramatic. But your way is more correct, well done.

    (2) “You’re comparing 9 games (correctly should have been 12) lost in 8/9th inning with 15 BS by Putz and K-Rod. This is not correct, since some of those 15 BS by Putz/K-rod resulted in wins by Seattle/Anaheim.”

    The blown saves by Putz and K-Rod that resulted in wins are of no consequence to our study, because we’re talking about the Mets. Just because Seattle and Anaheim were able to overcome the blown saves by their closers, does not necessarily mean that the Mets would have done the same. The point is that all of the blown saves by K-Rod/Putz came in the 8th or 9th, which are the innings they need to cover in ’09.

    (3) I disagree somewhat, going along with the same thinking as the “loss column” — you can always win more games, but you can’t lose less games. The percentage of games won after blown saves in the 8th/9th (but mostly 9th) is much lower than a blown save earlier in the game. But you’ve led into another issue, which ties into my argument — the main reason K-Rod saved 62 games was he had 69 opportunities. He may not get as many opps if the Mets can’t take a lead past the 7th, or if Putz blows a number of leads.

    Regarding Putz’s “subpar efficiency” … are you suggesting last year was an anomaly for him? If so then it’s fair to suggest that 2008 was an anomaly for both Wagner and Heilman, and call it a wash.

    Otherwise, I’m with you.

    (4) True that BS don’t tell the whole story, which was why I wrote this post. Good point on the ties, which “ties” into my argument — the Mets didn’t HAVE leads, or had very slim leads, going into the late innings during last season. Again, the fact the Mets have Putz for the 8th and K-Rod for the 9th means nothing unless the Mets are ahead by the end of the game. BTW why not find out for sure how many ties Heilman blew? Inquiring minds want to know!

  3. upson March 26, 2009 at 1:18 am
    Joe… good points although I really disagree on (2). [I agree that whether or not a game was won is not relevant when counting blown saves by K-Rod and Putz. However, under this assumption, it is also not relevant whether or not the Mets won the game after blowing a lead in 8th or 9th. Hence, it only makes sense to compare # of BS by Putz/K-rod and # of BS by Mets in 8th/9th — and this actually turns out to be 17, not 15, as I found some mistakes in my records.]

    Concerning other issues, I realized that the best way to settle this is to take a look on actual data. And the results are really worth it.

    Here’s my methodology… For each (late) inning, I counted the number of games in which the Mets pitcher entered the relevant half-inning in a “high-leverage” situation – i.e. when the Mets were in a lead of no more than 3 runs (=save opportunity) or when the score was tied. Then, I counted the number of games in which this lead or tie was blown in that particular inning. Here are the results (BS stands for blown save, BT stands for blown tie, the information in [ ] gives the distribution of sitations under which the half-inning started from Mets’ viewpoint: leads by 3+ runs / close leads / ties / any losing situation):

    ROAD games:
    8th inning: BS: 5/23, BT: 2/9 [20 / 23 / 9 / 29]
    9th inning: BS: 4/20, BT (= walkoff loss!): 4/9 [21 / 20 / 9 / 31]
    extra inns: BS: 0/4, BT (= walkoff loss!): 5/13
    total road games: BS: 9/47, BT 11/31

    HOME games:
    8th inning: BS 3/27, BT: 3/7 [19 / 27 / 7 /28]
    9th inning: BS 7/27, BT: 0/6 [20 / 27 / 6 /28]
    extra inns: BS 0/0, BT: 5/25
    total home BS 10/54, BT: 8/38

    Once again, I am only considering innings when the Mets had a close lead or were tied at the time when the first Mets pitcher entered the half inning. I’m not considering situation when the Mets had comfortable lead or when they were down. To illustrate this, the full distribution of situations facing the first Mets’ pitcher entering the bottom of the 8th inning on the road was: 20x lead by 3+, 23x close lead, 9x ties, 29x losing by any number of runs — this is captured in [ ] above.

    In total, there were 170 such high-leverage innings. Mets’ pitchers as a team failed 38 times.
    Hence, the failure rate of Mets’ pitchers was 38/170 = 22.4%.

    [Considering only “save situations”, the failure rate was 19/101 = 18.8%, considering only “tied situations, it was 19/69 = 27.5%]

    Let’s compare this with K-rod and Putz:

    K-Rod: BS: 7/69, BT: 1/5, failure rate in high-leverage situations: 8/74 = 10.8 %
    Putz: BS: 8/23, BT: 2/9, failure rate in high-leverage situations: 10/32 = 31.2 %

    Combined 18/106 = 17%, considering only “save situations” 15/92 = 16.3%]

    Funny thing is that now everybody can see whatever he wants in these data: You can very well claim that using K-rod and Putz during those 101 save situations last year would result only in a marginally better record – 2 or 3 fewer failures. However, one can also argue that Putz and K-rod would pitch also many of those “tied” innings where the Mets’ failure rate was higher. This would result in at least 4 or 5 fewer failures. And, as you pointed out earlier, late mistakes are hard to recover from. Hence, 4-5 fewer mistakes means at least 2-3 fewer losses and this would mean all the difference.

    Finally, I do not quite agree with your take on Putz’s “subpar efficiency”. You’re right that Heilman and Wagner did not have a good year. However, I’m sure the Mets acquired Putz under the assumption that he would be better next year. Otherwise, they would not have pulled the trigger.

    My prediction: K-Rod and Putz, barring injuries, will combine for roughly 20 failures in 140 high-leverage opportunities in 2009 – given the methodology above. Last year, that would mean 10 fewer failures which would mean roughly 7-8 more wins. Granted, healthy Wagner and Heilman could have prevented at least 5 failures last year as well.

    Disclaimer: I calculated all results by myself using Matlab. I’m pretty sure there are no mistakes but cannot guarantee anything. I downloaded Mets’ boxscores from retrosheet.org.

  4. upson March 26, 2009 at 2:01 am
    … and I forgot, Heilman in “tied” situations is actually quite fascinating. I found out the following: he entered 12 games under tied score in 8th inning or later. In 6 out of these 12 times, he “failed”. However, this does not tell the whole story as he pitched numerous innings on multiple occassions (most of them in extra innings!). Also, sometimes he entered an inning already in progress or left earlier.

    Taking this into consideration, he entered 21 distinct innings and “failed” only 6 times where I count for success 2 games where he entered the game and received the final out in a bases loaded situation, and I count for a failure when he left the inning with runners in corners and 0 outs.

    Hence, his failure rate given my methodology (adjusted for incomplete innings) was 6/21 = 28.6%. On the bright side, he was definitely Mets’ extra inning hero last season.

  5. isuzudude March 26, 2009 at 8:33 am
    I nominate upson as the smartest MetsToday poster. Were you up all night tabulating those numbers?

    But I thought the point of this post was to point out that Putz and Krod might not matter because it’s the guys between them and the starters that will cost the Mets some games this year.

    The more I think about this topic, the more I’m inclined to believe that it’s near impossible to predict how a team’s bullpen will perform. I don’t think many of us thought a group that included Chad Bradford, Aaron Heilman, Darren Oliver, Guillermo Mota, and Roberto Hernandez would have been as successful as they were in 2006, or that a group of many of the same faces would be so horrible the following year. Much of how a bullpen performs also is contingent on fatigue, injury, frequency of usage, and implementation of usage, all things impossible to foresee at this point in time. Still, a Putz and Krod tandem *should* be better than the Ayala/Heilman tagteam that cost the Mets the season in September, and the middle relief crew of Feliciano, Green, Stokes, Parnell, et al *can* be effective enough to hold the fort the majority of the time. I’m actually less worried about blowing saves now, but more worried about getting leads early in the game because of a makeshift back of the rotation and a questionable batting order. But now I’m just rambling.

    Is it April 6th yet?

  6. joe March 26, 2009 at 11:03 am
    Upson, fascinating research, thank you for taking the time and sharing. Great stuff.

    As ‘dude states, the main point is that although Mets leads should be ably protected by Putz/K-Rod, the bottom line remains that the Mets will need to hand them leads to protect — and I’m not sure the current menagerie of arms is up to the task from innings 1 through 7.

    As of today, Maine and Perez are bigger question marks than Livan Hernandez … that’s scary. The middle relief is questionable, as well, partially because we don’t know if any of the current relief candidates can be effective beyond one batter and partially because Jerry Manuel has proven inept as a bullpen manager.

    What worries me are those one-run and tie ballgames in the 6th and 7th, which Upson touches upon. If the Mets are tied, or down by one run in the seventh, who comes in to pitch? Will Manuel try to use a combination of Putz / K-Rod for multiple innings, to get the last 9 outs? If such a strategy works a few times, I guarantee we’ll see Putz trotting into the seventh on a regular basis, and 4- or 5-out saves becoming a regularity with K-Rod. Since K-Rod has always been exclusively a one-inning closer, we must wonder how such a scenario will play out, particularly in September. You know, that month the Mets perennially choke?