Opening Day and First-Pitch Tendencies

Many find that the easiest part of writing is the actual writing. Writing is just putting words onto a page. The more challenging part of writing is coming up with ideas for things to write about, and as far as baseball writing is concerned, that’s especially true around this time of year, when spring training is still going on despite near unanimous disapproval. So the baseball writer finds it particularly convenient when an idea is handed to him by somebody else. Tuesday afternoon, Brandon McCarthy presented an idea:

? for my stat enthusiast followers- in last 5 years, has any team’s opening day starter started the season with an off-speed pitch?

— Brandon McCarthy (@BMcCarthy32) March 19, 2013

Just like that, Twitter provides something pursuable. Thanks to PITCHf/x, Gameday, and MLB.tv, we can try to answer McCarthy’s question, and though the data gets less accurate the further back you go, that’s the data’s fault, not the user’s fault. Let’s see what’s out there.

First, consider what you might assume. Intuitively, you’d think the first pitch of a season would generally be a fastball, over the plate, taken for a strike. That just feels like it ought to be true — not for strategic reasons, but because of human nature. The pitcher wouldn’t want to throw a first-pitch ball, and he wouldn’t want to get cute before establishing the heater. The hitter wouldn’t want to risk making an out on the first pitch of the year, and humans tend to be risk-averse. It seems like Opening Day is the right recipe for a first-pitch called strike on a fastball.

So now we can look at what the data tells us. This past season, in 2012, nearly every Opening Day pitcher threw an obvious first-pitch fastball. Clayton Kershaw‘s first pitch was a little weird, but then, he was also sick. Justin Masterson‘s first pitch looked like a fastball with something taken off, as did Erik Bedard’s. The only guy we can’t be quite sure about is Carl Pavano. Here is Pavano’s first pitch, in the bottom of the first in Baltimore:

If I had to guess, I’d guess it was a fastball at less than 100%. I think it was more fastball than offspeed. However, it’s unclear, and the PITCHf/x data isn’t much help. Carl Pavano, then, gets a question mark.

And then there’s nothing but Opening Day first-pitch fastballs in 2011. Nothing but fastballs in 2010. Nothing but fastballs in 2009. In 2008, we’re missing some of the data, as PITCHf/x experienced a number of dropped pitches. But in 2008, we find something curious. Opening Day, between the Tigers and the Royals:

PITCHf/x caught Justin Verlander’s season-opening pitch at 84 miles per hour, labeling it a changeup. That pitch was followed by two dropped pitches, so we know something weird was going on that day, but here’s a potentially telling chart from Brooks Baseball:

It looks like Verlander started off with a changeup. That pitch was nowhere close to the fastballs he threw that day. But there’s still some mystery, on account of the PITCHf/x hiccups and the lack of MLB.tv video. This excerpt also doesn’t help:

1:10 p.m. ET: First pitch of the year is a fastball low, from Justin Verlander to David DeJesus. And first out is a groundout to the Tigers’ new first baseman, Carlos Guillen.

Was it a changeup? Was it a fastball that the PITCHf/x cameras misinterpreted? Because we can’t be absolutely sure, we can’t be absolutely sure that Justin Verlander threw the last Opening Day first-pitch non-fastball in 2008. There’s a certain likelihood, though, especially if you think Pavano threw a sinker in Baltimore. More generally, the point’s been established that the first pitch of a team’s season is almost always some sort of fastball. It feels natural, even if it’s entirely predictable.

We can keep exploring deeper. We’ve gone about as far as we can go into pitch types, but what about pitch quality and hitter behavior? We remarked earlier that we’d expect hitters to take the first pitch of a season. Let’s narrow our window to 2009-2012, leaving out the tricky and inconsistent 2008 data. That gives us 120 different Opening Day starts, which gives us 120 different Opening Day first pitches. Of these pitches, 45 were taken for balls. Meanwhile, 57 were taken for strikes, one was swung on and missed, four were fouled, and 13 were put in play. In other words, 85% of the pitches were taken, while 15% of the pitches were swung at. That is a very low swing rate, although it’s not exactly 0%.

What if we compare that against first pitches overall, instead of just first pitches on Opening Day? Thanks to excellent research by Matthew Carruth, I can tell you that over the window, 87% of all first pitches were taken, while 13% of all first pitches were swung at. This isn’t meaningfully different from the Opening Day data, meaning hitters don’t swing at more first pitches when it isn’t the first pitch of the season. Of note is that, on Opening Day, 38% of first pitches are balls. Overall, 39% of game first pitches are balls. This runs counter to what we might expect, with pitchers grooving the ball down the middle. They’ve still nibbled just about as much as usual.

Compare this data against first pitches for plate appearances, instead of whole games. The last few years, in all plate appearances, roughly 41% of first pitches have been balls. That’s awful close to the ball rate on game first pitches. However, in all plate appearances, roughly 26% of first pitches have been swung at. That’s nearly double the rate of game first pitches, suggesting that, indeed, hitters don’t want to make outs on the first pitches of games, but they don’t mind so much making outs on the first pitches of subsequent plate appearances. This seems to me like it’s suboptimal strategy, meaning it could potentially be exploited, but the gains would be very small, so, there you go. Maybe hitters just aren’t prepared to swing at the first pitch of a game. Maybe it’s important to see the pitcher throw a pitch before you decide to attack.

To sum up, a non-fastball to start a season is extraordinarily rare. There might have been one or two such pitches over the past five years. While most game first pitches are fastballs, evidence suggests they aren’t often grooved down the middle, like you might assume. Roughly two-fifths of the time, they’ve gone for called balls. Hitters already don’t swing very often at first pitches of plate appearances, but they swing even less often at first pitches of games, presumably because they don’t want to make instant outs. There’s no observable difference between swing rate at the first pitch of a season and swing rate at the first pitch of a normal game. We’ve got sample-size limitations, but we can still use what we have.

In a few weeks, there are going to be a lot more season-opening pitches. You’ll be paying closer attention, now, than you probably have in the past.