The Count Is King (Even After Accounting for Batter Skill)

Here’s a big, boring truism you surely don’t want to read an article about: as a pitcher, it’s better to be ahead in the count than behind. Good, great, fine. Thanks for the information, Ben, but let’s move along. We all know that, there’s no need to further prove it.

But wait! Here’s another truism that complicates the first one. Better hitters get ahead in the count more often. Mike Trout gets to 1-0 a lot more frequently than Billy Hamilton does — in roughly 48% of his plate appearances, as compared to a mere 36.8% for Hamilton.

So here’s a fact presented without context: major league hitters, as a whole, had a .363 wOBA after 1-0 counts and a .270 wOBA after 0-1 counts. Get ahead, hit better. But here’s some context, which at least slightly confuses the issue. The average wOBA of a batter reaching a 1-0 count was .322. In contrast, the average wOBA of 0-1 batters was .317. Better batters, in other words, really do reach advantageous counts more often. If you don’t account for that, you’ll probably end up over-valuing getting ahead in the count.

In fact, as a general rule, bad batters get to all of the worst counts more often. The best count in baseball, 3-0, also features the best batters on average. Some of the effect of a good count is exaggerated, the result of better hitters naturally getting to good counts more often:

Batter Skill By Count
After Batter wOBA
0-1 .317
1-0 .322
0-2 .314
1-1 .321
1-2 .318
2-0 .326
2-1 .324
2-2 .321
3-0 .329
3-1 .328
3-2 .325

But it’s a small effect! Counts of 0-2 and 3-0 have the most disparate batters. Just 15 points of wOBA separate the two of them. That’s meaningful — but quite frankly, it’s not all that big of a difference. It’s the same as the gap between Yasiel Puig and César Hernández in 2019. The difference between wOBA produced after 3-0 and after 0-2, on the other hand, is massive. After 3-0, hitters produced a .544 wOBA with a .725 OBP and .555 slugging percentage. After 0-2, they struck out 47% of the time and got on base at a .201 clip. That’s Babe Ruth vs. Bill Bergen, basically.

Just to drive the point home, take a look at the average wOBA of hitters in each count as compared to the outcomes they’ve produced after those counts:

Skill and Production By Count
After wOBA Produced Batter wOBA
0-1 .270 .317
1-0 .363 .322
0-2 .203 .314
1-1 .303 .321
1-2 .228 .318
2-0 .434 .326
2-1 .361 .324
2-2 .273 .321
3-0 .544 .329
3-1 .476 .328
3-2 .377 .325

Clearly, controlling for the quality of hitter in each count can’t explain the total variation within the count. This is — well, it’s not surprising, of course, but it’s still worth saying. You’d rather have Hernández (career .321 wOBA) in a 3-0 count than Puig (career .351 wOBA) in an 0-2 count. Their career lines in those counts are small samples, but Hernández hits a scorching .382/.783/.455 after 3-0 while Puig hits .156/.198/.236 after 0-2.

But just because count is much more important than batter skill doesn’t mean batter skill has no impact. Would you rather have Mike Trout after a 0-2 count or Austin Hedges after 2-1? Where is the line?

To try to come up with a better estimation of how much count affects hitters, I tried another tack. Instead of looking at the overall results of each count, I looked at how each hitter had done relative to their overall line. For example, if someone had a .400 wOBA overall and a .500 wOBA after 1-0 counts, their 1-0 skill was 1.25x (.500/.400). This isn’t a perfect process at the extremes — imagine a hitter who reaches a 3-0 count in every at-bat, and you’ll start to realize some of the problems — but as most hitters have similar-ish distributions, I find it to be an acceptable first step in approximating count-based adjustments.

For example, batters who face 1-0 counts produce lines roughly 13% better than their overall numbers in those plate appearances. In 0-1 counts, batters do roughly 16% worse than their overall lines. The whole grid looks like this:

wOBA Ratio By Count
After wOBA Ratio
0-1 0.840
1-0 1.13
0-2 0.63
1-1 0.94
1-2 0.70
2-0 1.35
2-1 1.13
2-2 0.85
3-0 1.73
3-1 1.47
3-2 1.17

A few quick methodological notes on the table: I took every pitch thrown in the majors along with the result of the plate appearance the pitch occurred during, then stripped out pitches that didn’t affect the count (fouls with two strikes). Then I just took every player’s performance after a given count, found their in-count wOBA and overall wOBA, and took a weighted average based on the number of appearances they made in that count.

How do I feel about these results? Eh, honestly I feel pretty mixed. The idea of using a multiplicative constant rather than an additive one is attractive to me because I, Ben Clemens, would have an expected 0 wOBA in an 0-0 count and also an expected 0 wOBA in a 1-0 count — adding 43 points of wOBA or whatever seems presumptuous. Additionally, it produced smaller squared error terms than a purely additive rule would have. But it gets weird around the edges, and for good reason. Mike Trout had a .436 wOBA last year. Put him in a 3-0 count, and this method gives him expected production worth more in wOBA than a walk, which is a bad prediction.

The best way to fix this is by getting a little more granular and working out projections for each outcome by count. That’s a better way of handling something heterogenous like wOBA, though it requires more careful handling. The pitcher has a say, too, something I ignored for the sake of this exercise — not a huge say, surely, given how little batter composition matters, but a say nonetheless.

And lastly, there’s one very interesting question I haven’t looked at yet: do some hitters have count-based “skill?” In other words, are there batters who are better at turning 1-0 counts into production, and conversely batters who don’t take enough advantage of getting ahead? Intuition says yes — but that intuition needs to be tested.

The point of bringing up these limitations and opportunities for further research isn’t to impugn the work I did here, though. Even without further refinement, one point is clear: the players who appear in each count matter a little, but they don’t come close to mattering enough to overcome the effects of count.

So the next time you’re watching a baseball game and see a 2-0 count to a banjo-hitting shortstop, adjust your mental model accordingly. José Iglesias, King Banjo himself, has a career 129 wRC+ after 2-0 counts, and a mark of 110 after 1-0 counts in a much larger sample. The hitter at the plate matters — but the count they find themselves in matters significantly more.

We hoped you liked reading The Count Is King (Even After Accounting for Batter Skill) by Ben Clemens!

Please support FanGraphs by becoming a member. We publish thousands of articles a year, host multiple podcasts, and have an ever growing database of baseball stats.

FanGraphs does not have a paywall. With your membership, we can continue to offer the content you've come to rely on and add to our unique baseball coverage.

Support FanGraphs




Ben is a contributor to FanGraphs. A lifelong Cardinals fan, he got his start writing for Viva El Birdos. He can be found on Twitter @_Ben_Clemens.

newest oldest most voted
Trey Baughn
Member
Member
Trey Baughn

Great article, and clearly establishes a baseline for a lot of interesting analysis. For example, it seems that if pitchers can improve their first pitch strike percentage even by small amounts, it could have a huge impact on their overall results (again, obvious but helpful).

cookthebob
Member
cookthebob

True. Interestingly though, 0-0 pitches have the smallest impact on PA outcome (measured by difference in resulting wOBA ratio on ball vs. strike). Pitches have a larger impact the later they are in the count, even excluding those that might result in a K or BB:
0-0: 0.29
0-1: 0.31
1-0: 0.41
1-1: 0.43
2-0: 0.60
2-1: 0.62