Phillies Bullpen Shockingly Good in August

The Achilles heel of the 2012 Phillies — sorry, Ryan Howard — has far and away been the bullpen. For most of the year, they have squandered leads large and small, sinking the Phillies to the bottom of the NL East. In May, the bullpen’s worst month, they posted a 5.23 ERA. Even after a few months of very slow progress, the Phillies still rank 12th in the National League in bullpen ERA at 4.42.

GM Ruben Amaro filled out his bullpen by signing Jonathan Papelbon to a four-year, $50 million contract and otherwise relying on cheap veterans and young pre-arbitration arms. Unfortunately, a few of the younger guys went down quickly, including Michael Stutes, David Herndon, and Justin De Fratus. Veteran Jose Contreras went down early as well, and the Chad Qualls signing did not pan out. Antonio Bastardo had some growing pains, punishing the Phillies for thinking he could handle the eighth inning with aplomb. Elsewhere, Michael Schwimer, Jake Diekman, B.J. Rosenberg, and July acquisition Josh Lindblom had their own struggles contributing in relief.

The past month, however, has seen a lot of progress among almost everybody. Their aggregate 3.79 ERA is a significant improvement for them, but more importantly, they had a 2.95 SIERA, third-best in the National League during the month. SIERA is an ERA retrodictor that looks at the factors a pitcher most controls — strikeouts, walks, ground and fly ball rates (all independent of defense) — to tell you what a pitcher’s underlying talent really is, on the same scale of ERA. So the bullpen’s 2.95 SIERA, compared to their 3.79 ERA, tells you that they were a lot better than their results indicated last month.

As a whole, the Phillies’ bullpen led the way in strikeout rate at 30 percent. The closest team to them was the San Diego Padres at 25.8 percent. Strikeouts are great and incredibly important to a pitcher’s success, but they had two issues that led to their demise: walks and home runs. The Phillies led the NL in walk rate at 10.6 and had the third-highest home run rate, averaging 1.1 per nine innings. However, the blame can be pinned almost entirely on Lindblom and Rosenberg. Lindblom allowed 11 walks and three home runs in 12.2 innings, while Rosenberg walked three and allowed two home runs in 8.1 innings. Combined, the two accounted for 44 percent of the walks and 56 percent of the home runs the bullpen allowed in August.

Individually, six relievers posted a SIERA below 3.75:

  • Antonio Bastardo: 1.55 (5.14 ERA) in 7 innings
  • Raul Valdez: 1.84 (0.93 ERA) in 9.2 innings
  • Jonathan Papelbon: 2.17 (2.84 ERA) in 14.1 innings
  • Michael Schwimer: 2.58 (7.04 ERA) in 7.2 innings
  • B.J. Rosenberg: 3.36 (4.52 ERA) in 8.1 innings
  • Jeremy Horst: 3.61 (3.90 ERA) in 11 innings

That’s without mentioning Phillippe Aumont, who joined the Phillies in late August. Four relievers — Valdes, Papelbon, Bastardo, Schwimer — had a strikeout-to-walk ratio of four or greater. Although Bastardo’s walk rate was high at 10 percent, he struck out one batter for nearly every two he faced. Comparatively, Aroldis Chapman‘s strikeout rate on the season is 47 percent. In other words, for the last month, Bastardo has been getting strikeouts at a rate comparable to the most dominant reliever in baseball this season.

September hasn’t started out as positively for the Phillies, especially with Sunday’s devastating loss. Still, the overall defense-independent stats on the season paint a very optimistic picture for the bullpen going forward. On the season, their 3.31 SIERA ranks seventh in all of baseball and fourth in the NL. A 2013 bullpen that includes Papelbon, Bastardo, Aumont, and a few more of the plethora of young arms the Phillies have at their disposal should make for some reliable late innings, a shocking concept after the relentless failures we have seen throughout this season. Whether through a direct decision or through circumstance, the Phillies have crafted for themselves a formidable bullpen that will have staying power for years to come.

Leave a Reply

*

29 comments

  1. Richard

    September 07, 2012 08:03 AM

    Yeah, I’m really encouraged by the bullpen going forward. Lots of live, young arms. And very nice to see Bastardo bringing it. I think he’s been the victim of bad luck, mostly, though also certainly his control let him down a bit.

    If guys like Diekman and Aumont can keep their walks down, the Phillies will really have something.

  2. Mike

    September 07, 2012 08:34 AM

    I have read a lot of references to bad luck on this site. Bad luck is when you have a runner on 1B and hit a deep ground-rule double in which the runner on 1B would have easily scored but has to go back to 3B. Bad luck is when you are about to field a ground ball and it hits a rock. Bad luck is when your team has a large early lead but the game gets rained out and the teams have to start a new game at a later date. Bad luck does not happen often. A pitcher’s inability to get batters out is not bad luck. At some point, you have to give credit to the other team.

  3. Richard

    September 07, 2012 09:00 AM

    You do not understand the references to bad luck, Mike. The bad luck I was referring to, other than just what it usually means (i.e., random variation, or sequencing), is the bad luck of pitching in front of a weak defense, especially a weak outfield defense. It is, by and large, a technical term.

    Bastardo’s propensity for giving up walks is not bad luck. His worse fortune this season on batted balls, in part, is. Though no doubt, if you are struggling with your control, you are more likely going to have to come in with something hittable. Even so, a better defense (and the Phillies defensive efficiency this season has not been good) can alleviate much of even those.

  4. Brendan

    September 07, 2012 09:05 AM

    Mike, this is not the site for you if you can’t get a grasp on the statistical concept of luck within the context of baseball.

  5. LTG

    September 07, 2012 09:32 AM

    The concept ‘luck’ in sabermetric analyses has not been refined enough to count as technical. It licenses two inferences which are in constant conjunction but obviously need not be by the lights of statistical analysis itself.

    The first inference is that we should expect the player’s performance to return to some normal baseline for that player (regression). The second inference is that the bad outcomes are not due to a bad performance by the player (innocence). But clearly there are cases (I bet many) where regression is licensed but innocence isn’t. Why? Because player’s are never perfectly consistent with their fine-grained performance aspects that are not duly capture by summative stats. The quality of a player’s performance tends to vary around a mean and vary in a more or less regular way. So, the player can temporarily perform badly (be not innocent) and give us reason to expect improvement (regress). So, ‘luck’ is not yet a coherent concept in baseball analysis in so far as it licenses these two inferences which often come apart.

    Two points of clarification. 1) This is not a veiled criticism of Cliff Lee this season. It is bad luck when all of your mistakes get hit hard. Even the best players get by on a margin of error. 2) I’m skeptical that we can distinguish good performance from luck on a fine-grained level (e.g., pitch by pitch or AB by AB).

  6. LTG

    September 07, 2012 09:34 AM

    Also,

    What BB said.

  7. Richard

    September 07, 2012 09:39 AM

    fair enough, LTG… I admit I was using the term “technical” somewhat loosely, but mainly in opposition to a simple sense of “dumb luck”.

  8. Mike

    September 07, 2012 10:23 AM

    I get regression to the mean, and I agree that when a player’s statistic is below his mean that we can expect improvement. That’s fine. The issue that I have is calling poor performance “bad luck” whether the term has been properly defined by sabermetricians or not. It implies that, through no fault of their own, the player’s poor performance is attributable to others or is otherwise out of their control – for example, a bad defense. Why can’t poor performance be the result of something the player himself is doing (e.g., tipping his pitches)? If all mistakes get hit hard (assuming a reasonable sample size), then I have a hard time calling that bad luck. I think the other team somehow knows what pitch is going to be thrown.

  9. Bill Baer

    September 07, 2012 10:32 AM

    Colin Wyers just posted an article at Baseball Prospectus that has some relevance to this discussion.

    www.baseballprospectus.com/article.php?articleid=18271

    The fact that something is indistinguishable from randomness given our data and our model does not make it random.

    [...]

    The reason we’re uncertain about Rosenthal’s theory is because we lack good ways to test it. We should, of course, try to seek out those ways and actually do that. But until then, his theory is of very little utility. It doesn’t help us predict what the Dodgers will do. (And let’s face it, the rest of the season is not so long that a poor performance out of the Dodgers would do much in confirming it.) And because the theory seeks to explain previously observed behavior, it has the uncomfortable feeling of being a post hoc rationalization, rather than predicting something (as J. Wheatley-Schaller of Vegas Watch pointed out, the article would be much more compelling had it been written BEFORE the poor run of performance).

  10. LTG

    September 07, 2012 10:44 AM

    I’ve never heard of a theory that wasn’t offered post hoc. A new theory is only a competitor if it does a better job of explaining known phenomena than previously extant theories. The only question is what predictions does the new theory make and will we discover them to be accurate.

    We can’t know what is in need of explanation until something has happened. The post-hoc worry is overplayed and smacks of impatience.

  11. Bill Baer

    September 07, 2012 10:46 AM

    I think he means that there were opportunities to publicize that theory but the person in question waited until things went wrong to present it.

  12. Phillie697

    September 07, 2012 10:52 AM

    @Mike,

    The simplest explanation can be made like this. When you flip a coin by definition it should have a 50% chance to land on head, and 50% chance to land on tail. The law of large numbers says that if you flip this coin 10,000 times, you are likely to get very close to a 50/50 distributions of heads and tails. This is elementary statistics that everyone can understand.

    But everyone ALSO understands that when you flip said coin just 10 times, there is a non-zero chance that you might get 10 heads, or 10 tails. In fact, the chance of getting EXACTLY 5 heads and 5 tails when you flip a coin 10 times is a shocking (to the average joe anyway) 25% only. In other words, you are going to get fluctuations when the sample size is not large. Most people, when flipping the coin 10 times and get, say, 8 heads, will call that luck, when in reality it’s simply the combination of small sample size and law of probability. But at the end of the day, you still realize the true chance of getting a head: 50%.

    This works the same in baseball. Pitchers have varying talents, and that talent can be translated into an expected performance, i.e. a reasonable approximation of what he SHOULD do on the field, given his talent and the things he can control. However, other things are out of his control, i.e. whether a fielder gets to a ball in play or not (you and I both know not all fielders are created equal), whether a ball hit in the air becomes a HR or not (you might argue that a pitcher should not “pitch” to power hitters, but statistic studies have shown that whether a ball hit in the air becomes a HR is something that the BATTER controls, not the pitcher so much; the best the pitcher can do is simply try to make the batter miss, hence why K’s are so important). You must remember that these pitchers aren’t pitching to idiots nor robots; the batters too have a varying degree of talent that the pitchers don’t have control over (trust me, if Roy Halladay can choose to pitch to Polacido Polanco 27 times a game, he would).

    So the best we can do is come up with an approximation of what we think the pitcher can do, i.e. the odds. However, in practice, just like a coin can land heads 8 out of 10 times, so can a pitcher’s performance deviate vastly from that approximation. If you would call 8 heads out of 10 luck, why wouldn’t you consider this luck as well? Pitchers aren’t gods; they can’t control everything. But just as you would still recognize that a coin that laded 8 times out of 10 on heads still has a 50% chance of landing heads, you should recognize that what happens on the field do not always correlates to pitcher’s true talent. As my 5 heads/5 tails example shows, to expect it to EXACTLY is simply just not smart.

  13. Dan

    September 07, 2012 11:12 AM

    I feel like the idea that a pitcher is “tipping pitches” is an over-used term to explain high home run rates coinciding with high strikeout rates. For instance, I remember the beat writers suggesting that Joe Blanton — who had the best K/BB ratio in the majors when he was traded, but also gave up the most homers — was tipping pitches.

    It seems to me that in a case like Blanton, or even Cliff Lee, it’s not so much tipping pitches as it is purely variation from the mean. Both pitchers have sharp stuff and good control; they live in and around the strike zone. So, while both pitchers are going to generate a lot of K’s, there are going to be plenty of times when the hitters’ barrels just run into a ball and smack it out of the park. This year, it’s happened more often than usual, and I think that’s gotta be chalked up to “bad luck.”

    As for the bullpen, I like Richard’s point that, if a pitcher is struggling with his control, he will have to come in the zone with something hittable more often than not. That certainly makes sense, and is a concept I’d like to see explored more (i.e., the value of getting ahead in counts).

  14. Phillie697

    September 07, 2012 11:20 AM

    @Dan,

    There HAS been studies done, and there is GREAT value in getting ahead in counts. The problem is, once a pitcher, assuming he has the control, start to throw a lot of strikes early in counts, hitters adjust. Again, like I stated above, pitchers aren’t pitching to imaginary robots who will do exactly the same thing every time, hitters adjust too, and pitching to Joey Bats 27 times in a game is vastly different than pitching to Freddy Galvis 27 times a game.

  15. Mike

    September 07, 2012 01:19 PM

    I get binomial distributions too (and other distributions as well). I don’t expect a .300 hitter to hit .300 every season, and I don’t expect a 3.00 ERA pitcher to have a 3.00 ERA every season. I understand fluctuations. I also get that sometimes things just don’t go your way on the baseball field that day (personally, I just have a real hard time saying “bad luck” in this context). Maybe the fielders don’t make the plays that day or maybe the umpire is squeezing the pitcher that day (tipping pitches is also an example that I previously used, but it’s certainly not the complete answer). Or, maybe the pitcher just didn’t prepare properly between appearances. Who knows. But, my original point is that it just seems to me that many people on this site use bad luck as the default explanation for poor performance.

    On the topic of control – Keep in mind that having control isn’t just the ability to throw strikes; it’s also the ability to throw balls out of the strike zone on purpose. Hitters do adjust, but so do pitchers.

  16. jauer

    September 07, 2012 01:57 PM

    If Joe Blanton is tipping pitches, then he must be the greatest pitcher of all-time given his strikeout rate.

  17. Richard

    September 07, 2012 02:31 PM

    “But, my original point is that it just seems to me that many people on this site use bad luck as the default explanation for poor performance.”

    We got that that was your original point, but since it’s actually not the case (despite numerous accusations to the contrary), and you’ve done nothing to demonstrate otherwise, the logical conclusion is that you don’t understand what people mean when they refer to luck.

  18. Phillie697

    September 07, 2012 02:45 PM

    @Mike,

    Of course you’re entitled to your opinions, but we here believe that if there is something fundamentally wrong with a pitcher in things that he can control (preparation, tipping pitches, etc), then it will show up in the numbers, i.e. K rate dropping (you’re not striking out anyone if they are hitting your pitches), BB rate rising (you’re not going to get the hitters to hit your crap that never goes into the strike zone), increased LD% and FB% (if you’re throwing straight shit in the middle of the strike zone, the hitters will hit you like a pinata). If you don’t agree or find troubling that sort of reasoning, then I can see why you have a hard time understanding why people call it luck.

    After all, when you say a pitcher performed poorly, you obviously based it on something, usually runs allowed or ERA. That directly correlates to what happened on the field (Ks, BBs, balls hit in play, what those hits are, what did the fielders do, etc). SIERA and xFIP attempt to isolate what happens on the field even further by calculating a pitcher’s “performance” based on what is deemed to be things he has actual control over, while normalizing everything else. If you have trouble accepting this, then voice your disagreement on what is specifically wrong with the methodology, then perhaps we can have a more substantive discussion.

  19. Bill Baer

    September 07, 2012 02:46 PM

    I can see where he’s coming from, but there’s a bit of a selection bias. If a player’s performance isn’t matching up with his results and it’s not due to bad luck, there is most likely a very obvious explanation that A) has been covered already and B) doesn’t involve much (if any) statistical analysis. For example, an injury.

    Many of the articles that I’ve written in the past (Cole Hamels’ 2009, J.A. Happ’s 2009, Carlos Ruiz’s 2010, Cliff Lee’s 2012, etc.) have dealt with luck because that was the explanation everybody else was missing, and needed the most elucidation. If luck was a better-understood explanation for player performance, it wouldn’t be such a dominant theme around here.

  20. jauer

    September 07, 2012 03:09 PM

    “If luck was a better-understood explanation for player performance, it wouldn’t be such a dominant theme around here.”

    Remember when people tried to accuse Cole Hamels of being disengaged in 2009? What’s more likely, poor BABIP/defense, or one of the top performers at his profession suddenly becoming disinterested?

  21. Mike

    September 07, 2012 04:18 PM

    Let me ask this a different way – This all started with my response to a seemingly innocuous comment: “I think he’s been the victim of bad luck” (in a reference to Bastardo).

    Let’s take his ERA (but I would be happy if you used any statistic of your choice like SIERA). His ERA was 2.64 in 2011. His ERA so far in 2012 is 4.74. His career ERA is 4.13. Can someone help me understand what the models say are the contributing factors to the difference in his ERA between 2011 and 2012 – that is, contributing to his poor performance in 2012. Is it 100% bad luck? If so, can you help me understand the contributing factors of bad luck? For example, bad defense was used previously. If you tell me that bad luck is mapped to bad defense and other things (just let me know what other things) then I will understand what you are saying. It’s just a mapping problem. I think “bad luck” in this context is a misleading term, but that’s just semantics. I can live with that semantics issue if I can understand what is fundamentally driving the phrase “bad luck.”

    In any case, I have no idea how SIERA is calculated so I’m trying to go a different route to understand Bastardo’s drop in performance, at least as measured by his ERA.

  22. Phillie697

    September 07, 2012 04:59 PM

    @Mike,

    The answer is exceedingly simple: He wasn’t nearly as good as his ERA suggested he was last year, and he’s not nearly as bad as his EAR suggests this year. You’re comparing the two opposite sides of his fluctuations and wondering what the heck is going on, but when in fact advanced statistics say he hasn’t changed much (SIERA/xFIP 2.93/3.56 in 2011, 2.61/3.39 in 2012).

  23. hk

    September 07, 2012 06:01 PM

    Mike,

    To take Phillie697′s comment further and put it into the context of luck, Bastardo was lucky last year and unlucky this year and his true talent is probably somewhere in the middle of last year’s 2.64 ERA / 3.30 FIP and this year’s 4.74 ERA / 3.75 FIP.

  24. Mike

    September 07, 2012 10:00 PM

    Interesting – So the “advanced statistics say he hasn’t changed much” (between 2011 and 2012), but the premise of this argument is that “he’s been the victim of bad luck.” And then I read later that “Bastardo was lucky last year and unlucky this year.” So I guess I’m confused – Is Bastardo the victim of bad luck in 2012 or is he about the same in 2012?

    More important, if Bastardo (or anyone else for that matter) is the victim of bad luck, can someone please tell me the components of the bad luck. It just seems strange to me that people can refer to bad luck as a given without actually being able to explain the sources of the bad luck. If you tell me that the reason for the difference is completely attributable to the Phillies defense or some other factors that the model determined, then at least that is an explanation for the difference. Otherwise, “bad luck” just seems like a completely empty phrase to me that is used in the situation in which “the player underperformed relative to where the model expected him to perform” – ergo, bad luck.

  25. LTG

    September 07, 2012 10:08 PM

    Well, I’m glad that’s settled.

  26. Phillie697

    September 10, 2012 11:57 AM

    @Mike,

    Bastardo has been a victim of HRs this year (which seems like every Phillie). His career HR% is 9.5%, and 8.4% last year, and this year it’s 13.7%.

    To put it into context, that’s only 2 “more” HRs he’s given up all season than his career numbers would suggest, but when you’ve only pitched 44.1 innings, two extra HRs tend to blow up your ERA. Also, he’s had a ridiculous 14.8% of infield hits, which I think even you can agree isn’t particularly something he has control over.

  27. Phillie697

    September 10, 2012 12:02 PM

    BTW, this underscores why reliever performance tend to fluctuate wildly year over year. The small number of innings they pitch tend to magnify fluctuations. 2 extra HRs one year, and bam, he’s a bad reliever. 2 less HRs the next year, and we start calling him the next coming of Mariano Rivera.

Next ArticleCrash Bag, Vol. 18: A Cacophony of Squeaky Octopi