No bullpen? Madness, you say. I’ll do you one better. We may not even need a starting rotation either. But I’m getting ahead of myself.
Like many around here, I've been scratching my head some about how to fix a bullpen that, arguably, has been the main factor in our recent playoff drought. True, worries had cooled down given the latest additions to the ‘pen, but I for one still have doubts. Apparently, I’m not alone in my puzzlement. In a recent Halolinks, WiHalofan confessed to being confused about his own feelings concerning the possibility of signing high-dollar Japanese closer, Kyuji Fujikawa. On the one hand, he knows that expensive relief pitchers are almost never worth the money; on the other hand, we need to frickin do something! Since the heyday of Percy, K-Rod, and Shields, our bullpens have cost us a LOT of games. Teams that go deep into the playoffs, always seem to have good to great bullpens, but building a good to great bullpen seems easier said than done. The problem, as I see it, is that relievers don’t pitch enough innings in a year to be statistically predictable. Any given season, a guy who pitches sixty innings could suck big time—Fernando Rodney over his years with the Angels—or completely kick ass—Fernando Rodney last year with the Rays. Luck, it seems, has more to do with bullpen success than most of us like to admit. In 2013, we have no way to predict whether Madson, Burnett, Downs, Frieri, or anyone else will shine or flop. We can work with certain probabilities, understanding that one given reliever is more or less likely to pitch well than another, but the margins of these probabilities are maddeningly slim. No one in baseball is a sure thing; relief pitchers are significantly less a sure thing than any other player. So what can we do, aside from what we are doing: signing as many lowish-cost relievers as we can manage, mix them with in-house options, and hope for the best? This is the sensible approach, but relying on hope just doesn't sit well with most of us. Surely there’s a better way to manage pitching, something either no one’s thought of yet, or no one’s been brave—or desperate—enough to put into practice. Remember, Moneyball was about using stats to find a novel way to build a baseball team—to take advantage of the current flaws in conventional baseball wisdom. The word is out on OBP; that’s done. But surely there are still plenty of fallacious and unexamined baseball assumptions out there, ready to be exploited by a visionary franchise hoping to get a leg up on the competition for a few years. Now, I’m no baseball whiz; I like the game, I watch it and read about it, but I’m not a math guy, and I quickly lose my way in discussions of advanced stats. I’m a writer of stories, better suited to science fiction than science, but I hope you’ll read on, if only to debunk or refine my ideas, half-baked though they may be.
Before launching into my little foray into the imagination, a bit of history, some consideration of current assumptions about pitchers, and a review of recently published research are in order. For a brief and interesting account of the history of pitching, check out this article by Baseball Prospectus’s Rany Jazayerli. What Jazayerli tells us, among other things, is that the current use of an array of relief pitchers supporting a five-man starting rotation began in the 1980s, and further specialization of relief pitching has been slowly growing ever since. What this suggests to me is that our present system has been in place longer than teams have been using advanced stats to aid in decision making. I have found little data supporting the idea that the current use of pitching is the most effective. Many of us bemoan Scioscia’s use of bullpen “roles”; the way he depends upon the same guy for the same inning no matter the circumstance strikes us as overly rigid, if not plain stupid. It often seems that our skipper turns off his brain in favor of a paint-by-numbers approach. Similarly, he frequently leaves floundering starters in way too long (I’m talking about you, Ervin), sometimes all-but surrendering the game by the third inning. We’d like him to loosen up, use more situational judgment, but can we really blame him for his strict adherence to doctrine? Isn't the whole system of how baseball manages pitchers—including the five-man rotation—equally rigid?
At the very least, baseball’s current use of pitchers is worth examining. In that spirit, I've begun to list and contemplate several baseball assumptions about the various roles of pitchers. This is not a complete or perhaps even a very good list, but it’s a start.
- Relief pitchers, especially closers, have “ice in their veins.” The idea is that relief pitchers are somehow different in temperament from starters—and that closers are even more different. They are cooler under pressure, which allows them to perform well in intense situations where other’s would succumb to nerves. I’m skeptical. Anyone with the cojones to pitch in a major league game without falling apart can do it in the late, middle, or early innings. If a guy can’t handle stress, he shouldn't be on the mound, no matter whether we call him a starter, closer, set-up man, or middle reliever. Furthermore, people learn to handle stress through practice. I concede that the ninth inning might feel a little hotter than any other, but I see no reason why only one guy out of twelve can handle that responsibility. If only closers can pitch the final inning, why are starters used so effectively as closers in playoff games? It’s because, I contend, they’re excellent and experienced pitchers, more important factors than some notion of temperament. I call this one a myth with a grain of truth thrown in.
- Some pitchers, especially hard throwers, are more suited to the bullpen because they lack the endurance required of a starter. Maybe, maybe not. I've found no statistical connection between throwing hard and durability. As for endurance, we humans tend to handle whatever we've trained for and what is expected of us. A pitcher expecting to throw 60 innings will prepare himself to throw 60 innings; my guess is that most relief pitchers could throw a lot more innings if required and properly trained, and if they can’t, are they really the kind of pitcher you want on your team? Some exceptions certainly apply, but again I’m skeptical.
- A guy with only one good pitch can get people out for one inning; a starter with four or five pitches can last a whole game. It’s always easier to pitch one inning than five or six, and this is true for guys with several pitches or guys with just one. I contend that the pitcher with a full arsenal is going to be better at getting outs no matter how many innings he pitches. Mariano Rivera aside, having a limited pitch arsenal is a weakness, no matter how many or few innings he pitches per game or per year.
- Pitching on a defined schedule puts starters at ease, and thus makes them more likely to succeed. Interesting idea. I've found no data to support it. If we’re just guessing, we can also posit that a pitcher who will potentially be called to the mound at any time most every game of the season will feel more involved with the team, thereby increasing focus and a sense of belonging, as opposed to zoning out four out of five games, twiddling his thumbs, sending tweets, or, worse, eating fried chicken and drinking beer in the clubhouse.
- Starters develop “a rhythm” as the game goes on. Statistically, this is untrue. Pitchers perform less effectively the more often batters see them—anecdotal evidence aside. Theoretically, the most effective way to manage pitching—if a team had enough pitchers at the same skill level—would be to bring in a new pitcher every trip through the lineup. This means five or so pitchers per game. Dave Cameron at Fangraphs actually suggests a four-man platoon strategy in this recent article. (This article, by the way, is one of the more radical calls for a reexamination of pitching management. Here’s a Sports Illustrated article from 2009 that argues for mixing up the five-man rotation by testing starters for durability and putting them on different schedules—in essence, everyone pitches as often as his body can manage. Other articles are more modest, arguing the relative merits of a four-man rotation on the one hand, a six-man on the other.)
What all this amounts to is that we have two classes of pitchers. Conventional wisdom has it that they are essentially different types of pitchers, one better suited for long stints, others better in short bursts. I submit that this whole foundational notion is based on various myths, and here lies the opportunity to take advantage of baseball’s general reliance on dogma over evidence. The real difference between “starters” and “relievers” is simpler: a matter of skill and talent. I placed quotes around the prevailing categories to suggest their arbitrary nature. We've invented these categories, and we can just as easily uninvent them. Fact is, aside from some notable exceptions, most relief pitchers enter the bullpen the same way. They are, for various reasons, demoted. This usually happens at some point in the minor leagues, and earning the job of closer is like being elected king of the homeless. Okay, maybe that’s overstating things. But I do think we can agree that starting pitchers are generally better at pitching than relievers. True, there is a gray area, where the best relievers may be better than some of the worse starters, but this doesn't disprove my general point. The only reason relievers seem better is that they are expected to pitch much, much less, and so they don’t suffer the same degradation of effectiveness as starters who must face the same batters several times in a game. This is also why they are so hard to predict. Baseball wisdom has it that a less-effective pitcher is good enough to get through a single inning, and sometimes that proves true, but if you think it through, you’ll see that we’re using this single inning requirement as an excuse to put in a relatively ineffective pitcher when better pitchers are sitting on the bench, or languishing in the minors, or are available through trades and free agency. Here’s my big question: wouldn't a team be better off dropping the categories of starter and reliever, in favor of evaluating pitchers only on one criteria: how likely they are to get batters out, regardless of the inning? Shouldn't a team put on its roster the twelve best pitchers available (sounds familiar, right?), and doesn't the artificial, self-fulfilling categorization of pitchers into two classes hinder attempts to acquire the twelve best pitchers? For example, let’s say there are fifteen starters available at some point and twenty relievers. Your team needs two starters and three relievers—for a total of five pitchers. For the sake of clarity, we’ll say all fifteen starters are objectively better than the twenty relievers. Wouldn't this team do better to sign their five pitchers—provided they can afford them, etc—all from the better, starter category? Ah, but what do to with this excess of starting pitchers? Well, that’s where a manager and pitching coach earn their money. They’d have to be very good at what they do if you take away the pre-established roles that our manager infamously over-relies upon. At this point, my idea gets a little fuzzier, a little more like science fiction than fiction. So be it.
Imagine a franchise—a bold owner, a smart GM—that decides to seriously rethink current notions of pitching management. They begin in the minor leagues, collecting data about some of the assumptions I listed above, compiling numbers and comparing them to historical averages. Let’s say they find what I’ve suggested they might find: that current pitching conventions are ineffective. They make a dramatic change, consisting of two parts. The first part of the plan has to do with personnel. In this franchise, from rookie ball to the MLB club, all pitchers know that there will be no distinction between starters and relievers. Individual pitchers may pitch more or less innings throughout a season, or placed more or less often in certain situations, given talent, health, and match ups. Each team will still have twelve pitchers, but they will be the best twelve the team can afford, and all of them will be expected to be able to pitch as often as required (or as often as their particular arms can take it). A few of those will no doubt be those pitchers traditionally considered relievers, but most, it stands to reason, would have been considered starters by the old reckoning. Let’s imagine a ratio of nine to three; that is, nine guys who have been trained to pitch several innings per game, and three who are accustomed to pitching one or two innings.
Now it comes to part two of the plan, coaching and management. On these teams, the manager and pitching coach have total discretion and total flexibility when it comes to their pitching staff. They can start anyone, anytime, and they can pull that pitcher at any time. Getting through three innings would be seen as a fine result, as long as the team is still in the game. The best pitchers will still pitch the most innings, but they don’t all necessarily have to come from six-inning starts. The guy who happens to pitch the first inning simply doesn't have his stuff that day? Pull him before the score is 6-0. Two on, nobody out in the ninth inning of a one-run game against the division rival? Put in the very best pitcher you have available at that time. Jered Weaver is rested enough to be effective for one out? All the better. This team, unlike the rest of MLB, tries to win every game—in the same manner current teams try to win every playoff game. All bets are off. Or maybe it makes more sense to aim to win every series, and perhaps the best probability of doing that would be, for example, to save the best pitching combinations for two games out of three. Example: the Angels are set to face the Yankees. CC is scheduled to pitch game one, Pettitte game two, and Kuroda game three. Out of the Halos’ twelve man pitching squad, dozens of combinations are available. It seems to me that with a very savvy coaching staff aided by badass probability software, a strategy can be devised that would ensure the highest likelihood of winning two out of three of those games, even if that means sacrificing a middling probability for, perhaps, the game that CC starts. Such a team, it seems to me, would have a distinct advantage over teams that use a rigid five-man rotation.
So how does it end? The team—let’s call them the California Angels (this is the future after all, and sometimes the future repeats the past)—wins the World Series that first year and dominates all of baseball for several years after. The GM is considered a genius. The coaching staff wins every award possible, and all of them except the manager are poached by other teams. Books are written, movies made, baseball wisdom is forever changed. Everyone lives happily ever after—all except the poor schmuck who wrote the humble fan post that first predicted the whole thing. His initial post, a poorly-written piece that received scant attention, was quickly forgotten by the time the plan was implemented, and the fan’s increasingly frantic attempts to claim credit resulted more and more often in insults and derision. Eventually he turned from irritating to scary, and the blog owners had no choice but to ban him. He died soon after, of a broken heart some say, and when his surviving relatives buried him (to quote Hawthorne), “they carved no hopeful verse upon his tombstone; for his dying hour was gloom.”
Silliness aside, I hope my ramblings spark some discussion about new ways to conceive of pitching. I’ll close with a not-particularly bold prediction. In twenty years, the management of pitchers in professional baseball will be radically different from current practice, and some team is going to profit for several years by leading the way. Wouldn't it be cool if that turned out to be our team?