clock menu more-arrow no yes

Filed under:

The magic of full counts

New, 30 comments

Thinking some more about baseball’s most intense count

Houston Astros v Los Angeles Angels of Anaheim
Even Mike Trout guesses wrong sometimes. This is a photo taken after he was rung up a called third strike during a full count.
Photo by Victor Decolongon/Getty Images

Here’s a fun question: what’s your favorite count in baseball? Given that you read the title of this post before clicking it, you can probably guess mine: the full count. This is probably not surprising — isn’t the full count everyone’s favorite count? When you were a kid, you didn’t recite to yourself,

World Series Game 7, bottom of the ninth, bases loaded, two outs, down by three, 2-1 count... the stretch, the pitch... and the batter takes a fastball on the corner and it’s now 2-2!

Right? Right? No, the full count is where the proverbial it is at. It’s not hard to see why: baseball is in many respects a game of rising anticipation, mixed with delayed gratification. The journey of each plate appearance through an accumulation of balls and strikes reaches a crescendo at “three and two,” at which point there’s nothing further to build to. At that point, on the next pitch, the valet of the house is going to announce that Mr. Rubber has arrived to pay a visit on Mr. Road (unless, of course, he’s waylaid by the ruffians of the Foul Ball Gang).

While reality is complicated, I like the elegance of the concept of the full count. (My predilection for elegance can be explained by what happens if you Google the adage, “First, we assume a can opener.”) It always brings back me to some time during the aughts, watching the Braves on TV (what a quaint idea these days...), with my doesn’t-know-what-baseball-is grandpa occasionally looking up from his book and stealing a glance at the screen. He’d watch for a few seconds, roughly the length of a pitch, and ask questions like:

  • “Bah, why doesn’t he swing at it?” Because it’s not in the right place
  • “So then, why isn’t the thrower putting it in the right place?” Because he wants the other guy to swing at it where he’s not going to be able to hit it
  • “But he guy with the stick isn’t swinging when he throws it in the wrong place!” Yep
  • “This game is ridiculous.”

Then a pitch would be thrown middle-middle, the batter would swing at miss, and my grandpa would harrumph: “So he was waiting for that the entire time, and then didn’t even hit it? Pfft.” Baseball. It’s a wonder it has any fans, I guess.

Without anecdotal abstraction to obscure my point, what I really mean is this: full counts are magical because they’re the ultimate risk-reward game-within-a-game that occurs on a baseball diamond. Broken down to its (overly simplistic) component pieces, the pitcher really has just two choices: throw a strike, or don’t throw a strike; the batter also has two choices: swing, or don’t swing. The pitcher wants the batter to not swing at strikes and swing at balls; the batter wants to swing at strikes and not swing at balls. In other counts, the stakes are way lower: both the pitcher and hitter generally get to survive to fight another day (or, really, fight again in the next twenty seconds). In a full count, Commander Adama’s mantra comes into play: you make choices, and then you live with them. As a pitcher, if you throw a strike, you have to live with it getting hit well (or getting a called strike three). If you throw a ball, you have to live with allowing a walk (or getting a swinging strikeout). As a batter, it’s the same: you can choose to leave the bat on your shoulder and live with a walk (or a strikeout), or swing away and hope it’s a strike that you can hit (or live with an ugly chase).

Depending on who you are, your brain may already be throwing up the roadblocks. But wait, it may be shushing, what about pitch mix? What about location? What about starting a pitch over the plate and having it dart outside? What about a back-door fastball? Moreover, this doesn’t even begin to get into whether a pitcher can even hit his spot well enough to throw a strike when he wants to. Yes, yes, I get all that, but in the end, the choice for the pitcher is the same. Either he’s going to try to throw a strike, or he’s going to try to get the opposing batter to chase. Everything else is just a variant of those two. It’s the same for a hitter: yes, he can look middle-in or try to spoil a breaking pitch away or anything else, but in the end, he’s either swinging or he isn’t. A decision is being made, at some point between the prior pitch and the current one reaching the plate, often in an infinitesimal, subconscious space. But it’s a decision nonetheless.

And, where there’s a decision made, there’s an opportunity to analyze it. I’m going to give you some facts.

  • Fact: in 2018, somewhat lower than half of all pitches were in the zone. Pitch tracking technologies and methodologies differ, but the various figures we have are: 49% (Statcast, rulebook zone); 43.0% (Fangraphs); and 48% (Pitch Info/Brooks Baseball).
  • Fact: in 2018, 47% of all pitches were swung at. (There’s no disagreement here, this is pretty immutable.) These are close, and we’d generally expect them to be — if they weren’t, my grandpa’s questions above would illustrate an even higher level of ridiculousness. Aside from considerations like batters always wanting to take a pitch for comfort, or because seeing pitches improves future performance against the same pitcher, we would expect these two numbers to be somewhat equal, so it’s not surprising at all that the swing rate is somewhat below the “in zone” rate from Statcast and Pitch Info.
  • Fact: in 2018, when a pitch was in the zone, it was swung at 66% of the time, and taken 34% of the time.
  • Fact: in 2018, when a pitch was out of the zone, it was swung at 28% of the time, and taken 72% of the time.

But, now, let’s talk about full counts, and only full counts.

  • Fact: in 2018, 58% of 3-2 pitches were in the zone.

I want to pause here. From the above, you know that full count pitches are more commonly thrown in the zone than other pitches. You also know that hitters, in broad strokes, swing more at strikes — they swing at strikes about twice as often as they don’t, and they take balls about three times as often as they swing at them. So, now, knowing this, what’s your guess for how often hitters swing at 3-2 pitches in the zone.

No, really, think about it before you scroll down. Hell, maybe scroll to the comments before scrolling back up and tell me what your prior is. I’m curious.


Okay, done? Well, here’s the answer:

  • Fact: in 2018, hitters swung at 91% of 3-2 pitches in the zone. 91%! That’s a lot!

Of course, hitters can’t know whether a pitch is going to be in the zone or not when they commit to swinging. If they could, it would be a very different game, and probably one my grandpa would find more appealing as background visuals. So, it’s unsurprising that, even though hitters swing only at 28% of all pitches outside the zone...

  • Fact: in 2018, hitters swung at 46% of 3-2 pitches outside the zone.

Going from “overall” to “full count,” hitter swing rates at pitches in the zone increased by about a third, but they increased by two thirds at pitches not in the zone.

Now, you’re reasonably well-informed with the above, but you’re missing one key bit of info: what actually happens after these decisions have been made? I’ve made a few diagrams that are hopefully interesting to look at.

If you were so inclined, you could start to use game theory to describe the overall batter-pitcher situation, but I’m going to skip right past that just for the sake of general comprehensibility. The bottom line is that pitchers and hitters are in a very unstable game of jockeying for position when it comes to this simple in/out swing/don’t swing framework. If you know nothing else about what the hitter is going to do, throwing it out of the zone makes sense: it transforms the opposing batter from Tyler Flowers to Dansby Swanson. But, of course, that comes with a risk: throw it out of the zone to a non-swinging hitter enough, and you’ve reaped the whirlwind of a .612 xwOBA thanks to walk after walk. Meanwhile, the potential gains of throwing it in the zone but not getting a swing, reducing opposing batters to Sean Newcomb’s batting line, are tempting.

For hitters, it’s the flip side of the same story. Don’t swing and you’re Mike Trout or two Ender Inciartes... at least until you morph into Newcomb because the ball was in the zone. Swing and the best you can do, on average is Johan Camargo, which is pretty great, but not Trout or Twoder Inciarte.

So, how does this general approach change when the full count is in play?

Things are crystallized. On all of these charts, blue is “worse than league average” and red is “better than league average.” You can see that on the first chart, who ends up in blue versus red is based on whether the hitter guessed right. But, on the second set of charts, the one about full counts, this isn’t realy the case anymore.

Note that these charts start with hitters being Freddie Freeman. Just by forcing a full count, a hitter has ascended from Flowers way upward. Similarly, by letting the situation get nearly out of hand, a pitcher has backed himself into a corner. (This is why the pitcher bubble starts blue, and the hitter bubble starts red.) But the good news for pitchers: of the four possible endpoint outcomes, three of them are better than league average. The only thing you can’t do is throw the pitch outside the zone and have the hitter not swing at it. The entirety of that possibility is what makes hitters start out as Freddie Freeman in full counts. Hitters have the opposite situation: they’re already Freddie Freeman when things start, but three of the four outcomes are against them. The saving grace: they can draw a walk.

I don’t know if this is what you expected. I’m not sure this is what I expected, though it makes sense upon examination. The most surprising thing is that upper-right hand corner outcome, the one labeled Ender Inciarte. Go back to the first set of charts. In all counts, a hitter that swings at a pitch in the zone results in an above-average rate of success, i.e., a 115ish wRC+ like what Johan Camargo posted last year. But, when restricting the sample to full counts, we get a below-average .303 xwOBA, akin to what Ender Inciarte did last year (and his 90ish wRC+).

What’s the takeaway from all this, why am I bothering? Well, it’s really simple. Probably too simple. Probably too obvious.

If you are a pitcher, throw a strike in a full count. Even if the hitter swings at it, his average outcome is actually not that great.

If you are a hitter, don’t swing in a full count. Even if you swing at a strike, your average outcome isn’t that great. Walks, meanwhile, are divine, and not that unlikely.

Of course, everyone already knows this. But that doesn’t mean they put this into practice. Remember those facts above? Here’s a rehash and a compilation.

  • Pitchers only throw something in the zone 58% of the time in a full count. Does that sound to you like following the bolded advice above?
  • Hitters swing at 72% of pitches in a full count. Does that sound to you like following the bolded advice above?

Again, none of this is that simple. As noted above, there are all sorts of considerations. This is all general. Maybe too general. It also doesn’t take into account aspects of game theory: the more a hitter is prone to take pitches in full counts, the less likely he will be to keep getting walked in full counts over time. Baseball, like life, is fluid, as everyone searches to find an edge. But right now, that edge seems to be fairly obvious. Pitchers, throw strikes. Hitters, take pitches. Those two will meet in the middle, one day (maybe). Until then, there seems to be a clear way to make the magic of full counts work for you.


It’s now been 2,000 words, and I haven’t mentioned a single player. So, let’s change that. In 2018, there were four hitters with xwOBAs above .500 on full counts: Jake Cave, Ryan Zimmerman, Adalberto Mondesi, and Andrew Knapp (minimum 20 full counts faced). The highest Brave? Ender Inciarte, at .426. Here’s the interesting thing about these guys: they aren’t uniformly guys that don’t swing. Cave swings more than average, especially at non-strikes, though Zimmerman was a particularly reticent swinger. Mondesi had a top-20 swing rate in baseball, though at least much of his was focused specifically on strikes. Knapp’s swing rate was pretty average-y overall, with a higher-than-average z-swing and a lower-than-average chase rate. And, we all know about Inciarte’s tendencies - he’s closer to Knapp than the other three guys listed in this paragraph.

(Because many full counts result in walks or strikeouts, the leaderboard of wOBA in full counts is similar to the leaderboard in xwOBA. Zimmerman, Cave, and Mondesi also appear right at the top. However, as far as Braves go, Tyler Flowers, Kurt Suzuki, and Nick Markakis all had wOBAs in full counts above .475 — they didn’t appear as high on the leaderboard above because they substantially outperformed their xwOBAs when making contact in those situations. Incidentally, Preston Tucker was the Braves’ worst full-count hitter by xwOBA at .265, but had one of the biggest xwOBA outperformances in those few situations on balls in play, leading to a fine .371 wOBA in full counts.)

What about players that were particularly poor in full counts? Two finished with xwOBAs below .200: Erik Kratz and Dixon Machado. There’s nothing special about Kratz other than him being a relatively poor hitter — he swings more than average at everything, but not in any egregious way. Machado, too, is a very poor hitter overall, but nothing jumps out about his take and swing rates, both of which are actually notably lower than league average. Both do make fairly weak contact — if one were so inclined, it could be interesting to see whether overall contact oomph is a potential explanation for poor full-count performance, preventing hitters from taking advantage of strikes when they do swing. The wOBA results aren’t too different from the xwOBA results for aforementioned reasons, but Braves fans won’t be surprised to find that Ryan Flaherty had one of the worst arrays of results in full counts in baseball: he managed to reach base in those situations (26 total PAs with six walks and three singles) nearly as many times as he struck out (eight), and of course, all three hits were singles, with one being a weak infield hit at that.

How about pitchers? If we look at the 166 pitchers with 50 or more full counts in 2018, we find that the two best pitchers at mitigating the full count were Adam Ottavino and Wade Davis, with xwOBAs below .250. In terms of starters, the next few names aren’t surprising: Zack Greinke (.258) and Chris Sale (.275), followed by Justin Verlander. After that, the field gets muddled: Jack Flaherty, Justin Anderson (who?), Matt Harvey, Heath Fillmyer. For example, Jacob deGrom is only 122nd out of 166 in mitigating xwOBA in full counts. The list of “worst pitchers by xwOBA in full counts” is much less recognizable, name-wise: Sal Romano, Homer Bailey, Matt Moore, Jordan Hicks, and Clayton Richard are the top five. In terms of Braves on this list, there are only four, though the order might surprise you. Julio Teheran allowed a .417 xwOBA on full counts, yet Sean Newcomb led the quarter with a .326 mark. These results are even better in actual results terms, as Teheran’s mark changes to .407 and Newcomb’s to a very nice .283. In fact, Newcomb is one of only nine pitchers with 100 or more full counts in 2018 to limit the damage to a wOBA below .300. Bet you didn’t see that one coming.


The above is fun from a name recognition perspective, but it really is just anecdata. What if we’re a little more rigorous? We already established the basic framework up above — pitchers should throw strikes in full counts; hitters should take pitches. How does that very (overly) simple framework explain success or failure?

I pulled together data on all batters with 20 or more full counts in 2018. (This cutoff was chosen because it’s roughly equivalent to the number of players with 200 or more PAs, which I generally use as my cutoff for “has enough of a sample to look at things that might be interesting.”) In case you needed further convincing that swing decisions in full counts matter, see the plot below.

Now, a correlation coefficient of 0.20 means that 20 percent of the variation in xwOBA on full counts can be explained just by swing rate. That intuitively makes sense: we wouldn’t expect it to be the only factor, but it’s a pretty big one. But part of the thing with these types of goodness-of-fit measures is that they take their job of identifying patterns seriously: close enough doesn’t cut it. Without getting overly technical, just consider this:

The pattern’s not perfectly smooth, and there’s some evidence that being overly passive in these situations might actually backfire (i.e., the pattern would be smoother if the first and second deciles had their average xwOBAs flipped). But, you more or less get the idea: (don’t) swing the bat (in a full count), meat.

Just for fun, the lowest swing rates in full count situations: Max Stassi (way ahead of everyone else, the only guy below 50%), Jesse Winker, Joey Gallo, Ian Happ, Franchy Cordero, Mike Trout, and Aaron Altherr. The highest: Dee Gordon (only player above 95%, Corey Dickerson, Eddie Rosario, Dixon Machado, and Orlando Arcia).

So, for hitters, we started with “hey yeah don’t swing in full counts” and that’s where we ended. How about pitchers?

I’ll be honest... that is not at all what I was expecting. Do deciles lend any more clarity?

I think the answer we’re looking for is “not really.” Instead, it just seems like there are clusters. If you throw too many non-strikes in a full count, it’s not going to go great for you. But pitchers are clustered together enough in the third decile and up that there doesn’t seem to be much difference or effect between slightly-variable strike rates and the outcomes they allow.

The ninth and tenth deciles are where things get really weird. Remember, these are the deciles with the highest rates of pitches thrown in the zone. The ninth decile is just a weird mix with no pattern (kind of like many of the deciles): it features Mets teammates Noah Syndergaard (.302 xwOBA-against in full counts) and Seth Lugo (.315) but also full count disasters Matt Moore (.469), Dylan Covey (.437), and Steven Matz (.424). A bunch of other guys in this decile I guess were too predictable in throwing strikes, including Bartolo Colon, Sean Manaea, Drew Pomeranz, and Tanner Roark, with xwOBAs-against in full counts also above .400. But then you get to the tenth decile, i.e., Guys Wot Don’t Walk Guys (kinda), and you have the really-good-in-full-counts Justin Verlander (.285) and Heath Fillmyer (.288), as well as the abominable Jason Vargas (.432) and the existing Ivan Nova (.404).

Is this just sample size, given that only about 160 pitchers had 50 or more full counts in 2018? What if we change it to 20, like the hitters? Well, here’s what happens. First, the correlation coefficient is still essentially zero. But, you at least do end up with this, which is somewhat nicer.

It looks like pitchers are perhaps just a little too idiosyncratic to summarize like hitters. I’ll let you draw your own conclusions about whether the advice from above still holds, that is, whether pitchers should try really hard to throw strikes in full counts or not.


The fun thing (for me, maybe not for you) is that the magic and drama of full counts doesn’t look like it’s going anyway. While reasoning could really go either way (more swings could either lead to fewer deep counts or more deep counts; worse pitcher command/control could do the same), the reality is that full counts have slightly increased over the last two decades, though they were largely stable up or only up a bit until the juiced ball era.

But, not to worry for those that like their baseball unchanging: this isn’t a dramatic shift. It’s really more akin to one extra full count a game, and one or two extra 3-2 pitches a game. But hey, the more opportunities to see a battle of do-or-die decisions at work, the better, I think.

So anyway, that was 3,500 words about my favorite count. What’s yours?