Which Pitchers Are Doubling Up to Start an At-Bat?

This is Nate Freiman’s fourth post as part of his August residency. Nate is a former MLB first baseman. He also played for Team Israel in the 2017 World Baseball Classic and spent time in the Atlantic and Mexican Leagues. He can be found on Twitter @natefreiman. His wife Amanda routinely beats him at golf. To read work by earlier residents, click here.

On June 7, 2013, I got the start against Chris Sale in Chicago. Roughly 22,000 people were there to see us beat the White Sox 4-3 on a Josh Donaldson sixth-inning grand slam.

I was on deck when Donaldson homered, and consequently faced a very angry Sale. He started me off with a slider. The pitch appeared to start more or less in the first-base dugout before catching the better part of the outside corner. Then he threw a changeup. I was geared up for 97. I buckled and took a second called strike. I was down 0-2 and still hadn’t seen the fastball. If you’re concerned about catching up to the fastball, the key is to slow down and think, “Be on time.” Hopefully that doesn’t translate to start a little early. That’s when you chase the back-foot slider.

Sale’s next pitch was 97 mph at the top of the zone. It looked even harder because I hadn’t seen the fastball. Strike three swinging. I got soft-soft-harded.

…

In my last post, I mentioned that at-bats are “path dependent,” meaning that each pitch is going to depend on the previous pitch. It’s nice to know what percentage of fastballs a guy throws. It’s really nice to have it broken down by count. Luckily there’s a really cool graphic for that on Baseball Savant. Here’s what it looks like for Blake Snell:

The chart shows that Snell throws 45.4% fastballs in 0-1 counts. In those counts, sometimes he got ahead with a fastball and sometimes he got ahead with offspeed. Do the pitches that came before it matter? Because soft-soft-hard is merely one example of a three-pitch sequence. I was curious whether MLB pitchers have measurable pitch-sequencing tendencies in other counts, too.

I filtered and grouped the 2018 Statcast data from Baseball Savant to look at approximately 52,000 plate appearances. In order to have a decent sample, I broke it down further between soft and hard. Usually, that’s about as much info as you can get in a game. When players relay pitches from the dugout or second base (just kidding, that never happens), oftentimes it’s just fastball or offspeed. In my experience, four-seam fastballs, two-seam fastballs, sinkers, and cutters all counted as fastballs. (There is a fine line between a cutter and a slider in the pitch data, so that is something to keep in mind.) Everything else was offspeed.

I looked at a pitcher’s fastball percentage when the previous pitch was a fastball and when the previous pitch was offspeed. Then I divided them to get a pitcher’s Ratio. A Ratio of 1.0 means that a pitcher has the same observed fastball rate regardless of the previous pitch.

This could have gone on forever, with nearly endless permutations, but I limited it to the most common two-pitch sequence a hitter sees: the first two pitches of an at-bat.

Let’s start by looking at that 0-1 count. A painted fastball for a first-pitch strike is a very different thing than a get-me-over breaking ball. As a batter, it’s incredibly frustrating when pitchers execute first-pitch offspeed for a strike, because it forces one into an 0-1 without having seen the fastball. The one has to rely on whatever info was gained while standing over in the on-deck circle.

Some pitchers appear to have no difference in what they throw based on the previous pitch.

Some pitchers, though, throw noticeably more fastballs after they get ahead with a fastball.

Justin Verlander, Junior Guerra: they’re following fastballs with even more fastballs.

Other pitchers, meanwhile, have the the tendency.

There’s Blake Snell again. It turns out that while Snell throws 45.4% fastballs overall in 0-1 counts, the times he gets ahead with an offspeed pitch, he’s significantly more likely to come back with a fastball. It’s also interesting that the top-five pitchers on this leaderboard have ratios much closer to 1.0 than the first group.

So those are 0-1 counts. How about at-bats that start with a ball? When the pitcher misses with a fastball 0-0, the hitter not only has the advantage of having gotten ahead in the count, but also (hopefully) of having gotten his timing down. Pitchers: can you execute a 1-0 right-on-right changeup? It’s a valuable skill at the upper levels to command a 1-0 offspeed offering after missing with the fastball. But what if the pitcher missed with first-pitch offspeed? I used to sit dead red and look middle-middle. The fastball was coming, or at least so I thought.

Here are the pitchers who have no apparent sequencing tendencies in 1-0 counts.

Here are the pitchers who have pronounced splits.

The pitchers on this list — most notably Rick Porcello — often double up their first pitch if they go 1-0.

There are also pitchers who tend to throw the opposite of whatever they fell behind with, though this tendency is much weaker.

As we saw with 0-1 counts, these ratios approach one much more quickly than with pitchers who double up.

Now let’s look at the entire league.

The chart above largely confirms what we saw in the leaderboards. As a whole, pitchers are close to neutral in 0-1 counts, though there are some with more pronounced tendencies. The 1-0 curve is noticeably shifted to the right, suggesting that the league leans towards doubling up after missing with a first-pitch fastball. I also looked at the effect of adding 1-1 counts to the sample. There’s more variation once you look deeper into an at-bat.

Is there a best way to sequence early in the count? I looked at both Pearson and Spearman correlations between the Ratio and both ERA and FIP and found nothing. These are tendencies that aren’t right or wrong, but are simply intrinsic to the pitcher.

These Ratios are observed and don’t control for matchups, defensive positioning, or situation. They also don’t take into account whether the first pitch was a called or swinging strike or a foul ball. We should also keep in mind that pitch-sequencing isn’t random. Consecutive pitches are dependent events. If the situation calls to start a hitter off with a fastball, there’s a good chance it would call for another.

There is a lot of information to go over before a series. Tendencies, while certainly useful, are still just percentages. And anyway, it’s possible to know what’s coming and still — if the pitcher executes his pitch — have a hard time hitting it. If I were a hitting coach, I might distill this entire report to a line or two in an advance meeting. Likes to double up. Likes to mix the first two pitches. Now go hit a double.