June 9, 2010
What Strasburg Threw
Heading into Tuesday night’s game, it seemed impossible for Stephen Strasburg to match the hype. MLB.com was offering free streaming video of the first four innings—but only while Strasburg pitched. When the Nationals batted, MLB.com sat in silence with a graphic telling us to await Strasburg’s return to the mound. The official Twitter account for Major League Baseball announced that the lineup card was being sent to Cooperstown, to be preserved. (Other game memorabilia, on the other hand, would be available for auction, @MLB tweeted.)
So there was almost no way for the game to live up to the hype. Well, except for 14 strikeouts and no walks in seven innings. That would probably be a way. And that’s exactly what Strasburg did as he pitched the Nationals to a 5-2 victory over the Pirates.
At least in this start, Strasburg was at the intersection of having good stuff and getting good results out of it. I’m not a fortune teller—I can’t tell you what will happen. But let’s look at the stuff he was throwing that game and see what we can find, shall we?
Actually, some notes on his two-seamer. If you look at how he grips the ball, his grip is slightly offset from a traditional two-seam grip. Mike Fast likes to call this the one-seam fastball. He still gets more “sink” from the ball than if he were throwing his four-seamer, but the orientation of the seams as the ball approaches the hitter will be different.
Some descriptors of the average characteristics of the pitches, after I grouped them together using a k-means algorithm (rather than using the stock classifications provided by Gameday):
Speed is measured by the Pitch-f/x system at the so-called “release” point, or about 50 feet from the back of home plate. Horizontal and vertical movement are measured at 40 feet from the back of the home plate, in comparison to a hypothetical pitch affected only by gravity.]
His go-to pitch was usually the four-seam fastball, followed by the curveball. (We’ll look a bit more at his pitch selection in a moment, so we’ll just pass that for now.)
We can also look at similar data for each pitch, graphically:
Red represents the four-seam fastball, blue the two-seamer, green the changeup, and black the curveball.
Speed is the same as reported above—speed at 50 feet from the back of home plate. Spin represents the axis of the spinning baseball in flight, which controls the direction of the horizontal and vertical movement of the pitched ball.
He gets a lot of break on his curveball—it’s rather sickening, really. And he’s got a massive gap in speed between that fastball (it passed the 100-mph mark just once in the start, according to F/X) and that curveball, which lives in the low 80s.
Now let’s look at where Strasburg was releasing the ball. All location graphs are going to be from the perspective of the catcher, facing out from the backstop toward the pitcher:
You can tell he’s a right-handed pitcher with a three-quarters delivery, at least. Let’s zoom in a bit:
For the most part you see a pretty tight grouping of release points here. People are going to be tempted to look at that graph (or similar graphs around the Internet) and say he’s releasing his curveball a bit higher than his other pitches, but he isn’t—or at least if he is, you can’t tell just by looking at that graph. The Pitch-f/x system is unable to distinguish between the movement of the pitcher and the ball at the actual point of release, so it picks up the flight of the ball a bit later in the trajectory and extrapolates back to 50 feet, or a few feet after the pitcher has actually released the ball.
In order to throw a curveball—with its pronounced downward break—for a strike, you have to give it more of an upward launch angle at release. By the time the F/X system starts reporting the flight of the pitched ball, that increased “loft” under the curveball means it’s a bit higher up than the other pitches a pitcher throws. It doesn’t mean his release point for that pitch is any different, though.
Now let’s look at what pitches he was throwing and where he was throwing them, looking only at right-handed hitters:
The strike zone presented is based upon the average reported top and bottom of the strike zone for the hitters Strasburg faced—Pitch-f/x operators record the top of the zone around the belt, as it’s typically called, rather than the rulebook zone at the letters.
What I most want to call you attention to is his curveball—look at it! In the zone! What’s great about Strasburg’s curve isn’t just how much it breaks, but how he can throw it for strikes. It’s disgustingly good. Also notice that against right-handed hitters he seems to really favor the four-seamer over the two-seamer; and while he’ll pepper the four-seamer all over the place, he keeps the two-seamer in to the hitter.
Now, for left-handed hitters:
You see a lot more two-seamers this time around, and this time out away from the batter. (Although it’s the batter that’s moving—he’s throwing the two-seamer at the same place.) And again, can I just say—look at that curve being thrown for strikes? It simply isn’t fair, is what that is.
There’s also that one errant two-seamer—well, I’m calling it a two-seamer. I’m guessing it’s a pitch that just got away from Strasburg and went high. Unless someone actually asks him and he can tell us what grip he was using there, we’re probably never going to know what he was actually trying to throw with that pitch.
Need for Speed
So, how did Strasburg’s speed change over the course of the game? Looking only at four-seam fastballs:
I know some people said it looked like he picked a little extra “oomph” at the end, and he did a bit, compared to where he was at right before, but you still see a slight downward trend overall. To the extent you want to attribute it to anything, I’d say that Strasburg knew he was facing his last batter and decided to put a little extra on it, not that he was getting stronger as the game went on.
Notes and Asides
For those wondering, all the graphs (except for the speed over time line graphs, which was done in Excel) and pitch classifications were done using GNU R.
To classify the pitches, I used a k-means clustering solution with four centers (that is to say, I knew he had four pitches) using speed, vertical movement and horizontal movement as the input variables. I then assigned the clusters labels based upon the characteristics of each group and what pitches I knew he was throwing.
Also, many thanks to the invaluable Mike Fast of the Hardball Times who chatted with me about the data while I was making the graphs for this article.