Ruminations on Pitching Efficiency

Derek Norris helped contribute to the A's league-leading walk rate. (via Keith Allison)

Derek Norris helped contribute to the A’s league-leading walk rate. (via Keith Allison)

We often encounter baseball colloquialisms related to rating players based on their perceived production. “That guy has a great eye at the plate” and “so-and-so plays great defense” are phrases we’ve all heard more than a few times when sitting in the stands. These eye tests are great because anyone can do them, but they’re not great because they’re usually wildly inaccurate. Today we’re going to address one of these time-honored eye tests and try to get to the actual value of something that may not currently have a way of being measured effectively.

This is, at its core, a conversation about judging pitching efficiency on both sides of the ball. I hope that attempting to conceive a better measurement for efficiency on the mound can open a dialogue on the subject and will lead to input from other baseball minds that may be able to build on the ideas I’ve laid out.

One area where we have a gap in our understanding of pitching success is how each at-bat influences pitcher fatigue and efficiency. One main strength of the recent franchises that are oriented to high on-base percentages is their propensity toward working deep into counts as hitters. We know these teams take walks (let’s use Oakland as an example — the A’s 9.4 percent  walk rate was tops in the majors this year). We know that walks are good for offense, because base runners lead to runs, and runs lead to wins. However, just for a moment, let’s take walks out of the equation and focus just on the desire for the offensive team to work into deep counts. The A’s work into deep counts because they want to increase walks, but they also do this for a completely different reason—one that forms part of a hitting approach: “working the pitcher.”

We already have a simple measurement for this phenomenon of the player at bat “working the pitcher”: pitches per plate appearance. Unfortunately, this doesn’t really tell us a tremendous amount, as it’s a statistic without much context. We have to be educated enough to know that more pitches per plate appearance is supposedly a good thing, and for X, Y and Z reasons. And, being more or less intelligent baseball fans, we can infer the main positive outcome: increased walks. As we would expect, there is a positive correlation between pitches per plate appearance and walk rate. Let’s look at that relationship for the 2014 season:

BB%-PA_Comparison

There’s a tangible, positive result to seeing more pitches. As you might also imagine, seeing more pitches leads to more deep counts, including counts with two strikes. This leads to a slight uptick in strikeout rate, but that correlation is a little bit sketchier due to some outlier teams that may be skewing the results. We can say with some degree of confidence that any negative results to seeing more pitches is minor for the batter over the long term, and definitely offset by the very strong and evident positive correlation between more pitches seen and more walks.

We can also identify some of the philosophical reasons for being patient at the plate outside of the positive outcome of drawing more walks. Driving up a pitcher’s pitch count leads to getting to the bullpen faster due to the soft ~100 pitch cap that most teams use for their starters; it may also turn the lineup over more times throughout the game, leading to more opportunities for hitters to bat. These things are difficult to quantify, but the implications of getting to the bullpen early in the first game of a four-game series against a division rival are obviously positive.

With all of these myriad ideas laid out, let’s focus on what we may be able to measure statistically. There need to be two perspectives to a new way of looking at this efficiency: one from the pitcher’s perspective and one from the hitter’s. Let’s talk about pitching to begin with.

Pitching

What is pitching efficiency? In terms of what we’re discussing, it’s fewer pitches, more outs. Pitching inefficiency is more pitches, fewer outs. With that in mind, we can deduce some pretty quick points on the subject by stripping away a few preconceived ideas built up over a lifetime of listening to play-by-play announcers. Let’s focus on the types of pitchers that may be more efficient.

First, and this is very simple: an out is an out. Strikeouts are sexy, and pitchers with a lot of them have been shown to generate weaker contact and give up fewer hits. Strikeouts are also usually inefficient. They require at least three pitches per plate appearance (usually more), and, while the pitcher gets to feel good about his stuff and preen after recording a strikeout (unquantifiable, to be sure), he didn’t help his pitch count trajectory above the average unless he threw fewer than four pitches, which is the average number of pitches per each major league plate appearance (the exact figure is about 3.83 P/PA for most recent years).

Second, ground balls and infield fly balls are good for efficiency. This has been covered on FanGraphs, but grounders and IFFB get turned into outs more often than fly balls and line drives. Without venturing too far into overstatement here, our prototypical “efficient pitcher” is one who can pitch to weak contact on the ground and in the air. When and if we do formulate a metric to measure pitching efficiency (more on this in a bit), we shouldn’t attempt to capture a pitcher’s true worth or entire skill set, along the lines of SIERA; we should simply try to establish how good a pitcher is at getting outs with as few pitches as possible compared to the rest of the league.

There is already one very simple and fun measurement for starting pitcher efficiency: the “Maddux,” envisioned by Jason Lukehart. If a starter tosses a complete game shutout with under 100 pitches, he can claim the vaunted moniker of the first-ballot hall of famer for his box score. While rare and brutally simplistic, the Maddux is a boiled-down encapsulation of what we’re looking for with a pitching efficiency metric – one that puts stock into quick and efficient outs, however they come.

With those two simple points and the “Maddux” in mind, we can dream of a most brutally efficient of all perfect games – 27 first-pitch outs on ground balls/infield flies. This would be the pinnacle of pitching efficiency. Yes, this relies completely on hitters swinging and making contact every first pitch, and yes, this is stunning hyperbole and will never happen. But it’s a fun thing to keep in mind, and it tries to cut through some of the biases that we have failed to quash entirely when watching a pitcher during a game. This doesn’t necessarily say strikeout pitchers are inefficient pitchers; conversely, ground ball pitchers aren’t always efficient.

In a general sense, we know that the best pitchers in baseball blend strikeout ability with the ability to generate weak contact against them. However, the bias we’re faced with when attempting to measure a pitcher’s efficiency is that baseball as an institution (and consequently our own commonly-held beliefs) has put more face value on strikeouts, and that’s not entirely correct. Let’s keep that in mind moving forward during this conversation.

For pitching, we should attempt to establish a baseline “average pitching efficiency game,” much like a quality start, to use as a point from which we can judge a pitcher’s performance as above or below league average. To do that, we need to create a formula that compares a starting pitcher’s total pitch count to league average pitches per plate appearance, multiplying that latter figure by the number of batters faced during the start to compare how many pitches a “league average” starter would throw. In this current theoretical formula, we could assign a positive value for allowing an earned base runner (a hit, walk or hit by pitch), and a negative value for an out. Since we would want this to be a fielding independent statistic, we would also assign a fielding value modifier depending on which team the pitcher plays for, much like FIP metrics.

Pretty is What Changes
Take notice, baseball. Few things endure just as they've always been, casual pursuits least of all.

As always, more questions arise on our journey toward finding an answer to a question. For the pitching efficiency metric, these questions mostly have to do with the proper weighting of any potential positive or negative modifiers – such as what the value of an out should be over the value of an earned base runner. These are big, theoretical questions. A formula has no purpose if the number spit out at the end is without context or meaning. Giving these potential positive and negative modifiers proper weights is more than I can accomplish in this article, but I believe this is the right path, and these are the right questions to ask.

This presents us with the crux of the issue. In a larger sense, how exactly should we measure pitching efficiency when/if creating a new metric? Should we measure by the number of pitches per plate appearance with the ratio of earned base runners to outs compared to league average, or by some other factors? How do we differentiate between starting pitchers and relievers?

Hitting

From the hitting perspective, the most interesting question concerning pitches per plate appearance is what the definition of a productive out should be. When a batter has an 11-pitch at-bat during the first inning against a great starting pitcher and then strikes out on the 12th pitch, is there still no positive outcome to that at-bat? In an immediate sense, there isn’t; maybe the inning ended with the strikeout and there were no runs scored in the frame.

In a larger sense, there certainly was a positive outcome. Especially against top-tier starting pitching, a long at-bat is inherently positive regardless of the final outcome. In the modern game of capped pitch counts and OBP-centric offensive approaches, long at-bats aren’t a pleasant surprise anymore – they’re part of franchise mindsets. If teams are actively viewing long at-bats as a positive phenomenon, we should attempt to quantify them.

Creating a metric that quantifies a hitter’s effect on a pitcher’s pitch count comes up against some of the same issues we ran into with the pitching metric: How much weight do we give to outs vs. hits, and do we factor in runs? The list of questions runs on. Still, the offensive aspect of this question is incredibly fascinating simply due to this revised idea of a productive out. Do we increase the valuation of league-average players that have an above-average ability to foul pitches off and keep at-bats alive, even if they result in outs at an average rate?

Advanced metrics are, at their best, a way of cutting out the subjectivity inherent to eye tests and getting to the root of both what production is and what it could be in the future. Examining this question tells us that there are gaps in our knowledge of cause and effect for pitching efficiency on both sides of the ball. There is value for pitchers and hitters that we currently aren’t measuring with the metrics we have. The sabermetrics community has done a great job of covering most parts of the game, but there are areas that don’t yet have metrics to fully explain them, and there always will be. There will always be another metric to be created, just as there will always be another metric to be honed and improved upon. With the journey above, we can say that trying to create new metrics yields questions, frustrations, and illuminations – some of which turn into hits, and some of which turn into productive outs.


Owen Watson writes for FanGraphs and The Hardball Times. Follow him on Twitter @ohwatson.
newest oldest most voted
Ben C.
Guest
Ben C.

One small quibble– “Second, ground balls and infield fly balls are good for efficiency. This has been covered on FanGraphs, but grounders and IFFB get turned into outs more often than fly balls and line drives”

I believe grounders actually go for hits more often than fly balls. Their potency is mitigated, however, by the fact that they rarely go for extra bases and virtually never result in home runs. On a runs-per-out basis, grounders are better for pitchers (but not because they get turned into outs “more often” than fly balls).

Rich M.
Guest
Rich M.

Pitches per Earned Run (PPER)

GreenMountainBoy
Guest
GreenMountainBoy

Total Bases + BB/Total Pitches