Analyzing Gerrit Cole's dominant month of September

There were few starting pitchers better than Gerrit Cole down the stretch. In fact, if you want to be precise, only six other pitchers mustered a higher fWAR than Cole's 1.3 mark in September. It's probably just easier if I put some of the main numbers in table format with the September ranks. It went something like this:

Pretty impressive indeed. Now, here comes the inevitable question: How? What was different about September?

Obviously, a question that broad comes with a few different layers. Layer No. 1 is pretty simple: experience. Cole debuted in early June and was a major part of the Pirates playoff push, so, essentially, he learned on the job. He didn't lack talent, but hey, experience is experience, and it led to a September breakthrough. It's not something we can quantify through numbers, or at least accurately quantify it. It's just purely common sense, which, in this case, will have to do.

As for the things that are quantifiable through numbers, there are a few of them, most of which stem from Cole's pitch usage. So we'll start there.

Right off the bat, a few things pop out. One, Cole backed off his sinker in September. Two, he was essentially all hard stuff in June as he got his feet wet (sinker counts). Three, the fastball was consistently a common fixture in his arsenal. And the big one: the curveball became a more frequently used weapon. More on that in a bit.

The infamous sample size caveat isn't much of an issue here. Cole debuted on June 11 and proceeded to earn three additional June starts, and from there, he started five games in July, five in August and five in September. So no, the numbers aren't skewed by a tiny sample size, if that was a concern.

The only plausible scenario in which one could quibble about sample sizes would lie in Cole's pitch counts, which, unsurprisingly, weren't too high. And that makes sense. I'm going to go out on a limb and guess that the Pirates wanted to ease Cole into the job, like any other team would with their coveted pitching prospect, and they kept him in the 90-pitch range for nearly two months, until he finally topped the 100-pitch plateau in early August.

The point, of course, is that Pittsburgh gave Cole a bit more leeway as the year went on. Since more leeway equals more pitches per start, there's a gap between the overall pitches he threw in June and the overall pitches he threw in September. Or August. It's not big. Going of some simple addition, Cole threw a total of 340 pitches in June. In September, he threw 472, with the middle numbers being 442 (July) and 453 (August). No, not the biggest discrepancy, but, it's something to note.

But since that's just a footnote, let's discuss Cole's curveball, because that's where we can start kicking the tires on some theories for his September dominance.

First, though, let's remember that the Uncle Charlie wasn't Cole's go-to weapon in the minors. It was actually his slider. Baseball Prospectus actually listed it as one of the Minor League's best sliders, and in 2013, that same slider yielded an opponents' .162 batting average, .108 ISO, and .213 BABIP. Those are three darn good numbers, but in comparison to his curve, they fall a bit short. With the same numbers used above, Cole's hook yielded a slash of .206 AVG, .064 ISO, .250 BABIP. And there's this tidbit: He didn't allow a hit on his curveball in September, per Brooks Baseball.

Basically, Cole finished 2013 sporting two solid off-speed pitches. We knew he had one (slider) when he arrived. And he by no means did he ditch it. Instead, he added a curveball, making him even more lethal. Besides the raw numbers, there are a few trends of note, with the most notable one being the increased nastiness of Cole's curveball.

Observe:

September is indeed the readily noticeable outlier Whiff%-wise, as Cole's curveball Whiff% nearly doubled from August to September.

As for how Cole almost doubled his curveball Whiff%, well, let's knock off the obvious: Increased vertical and horizontal movement wasn't a noticeable trend. Sure, if you want to nitpick, there was a tiny difference from August to September--about an inch of more vertical bite. Nothing dramatic, though. His horizontal and vertical release points also didn't stray from his norm.

That leads us back to the drawing board. I thought about some alternative explanations for Cole's nasty hook, and I wound up questioning if Cole started to throw his curveball outside of the strike zone more, specifically in the dirt. It just so happens that there's something to this.

Per some research by the Beyond the Box Score research team (numbers courtesy of PitchInfo), we found that Cole threw 82 dirtballs in September, by far the most for any month. That 82 total includes all pitches, not just curveballs. But if we narrow our search to curves, we'll arrive at 45 curveballs that hit dirt, or about 51%.

So what prompted Cole to suddenly start wielding more curveballs, and more specifically, more curveballs in the dirt? Well, that's the big unknown.

We can try to take an educated guess, or guesses. One possible theory: In 2013, Pirates catcher Russell Martin called the fewest amount of curveballs among all catchers who caught at least 10,000 pitches, per Baseball Savant. Naturally, Pirates pitchers threw the fewest amount of hooks too.

But, there's more to it. Yes, Martin, for some reason, was unwilling to call curveballs, but the raw data doesn't consider the tendencies of Pirates pitchers. Because, you know, if a pitcher boasts a good curve, he's going to use it, regardless of a catcher's tendencies. Vice versa for a pitcher with a mediocre to bad hook. Basically, the catcher only decides how to best use a pitcher's arsenal.

A good example: A.J. Burnett, who, by FanGraphs' pitch values, had baseball's best knuckle-curve in 2013. And if you wanted to get specific, he'd have baseball's sixth-best overall curve if you combined his wCU/C and wKC/C. Safe to say: Martin's pitch-calling tendencies didn't stop Burnett from using his knuckle-curve with great frequency (how about major league-leading frequency).

So yeah, this theory doesn't help much. Perhaps Martin specifically wanted Cole to find the strike zone with pitches he was comfortable with before he consistently started to break out the curve. Really, anything is purely speculation until we can delve into Cole's mind--which, obviously, isn't going to happen.

What isn't speculation, however, is the ripple effect. Batters often went fishing for Cole's curves in the dirt, causing him to use it more and ultimately go on to muster a 31.2 K% in September--no other month came close to matching that mark. And we already went over his impressive numbers outside of K%. Whatever Cole did, and I think we now have a good idea as to what that is, worked. How will an increased usage of Cole's curveball fare over a full season? We'll have to play the wait-and-see game to get an accurate answer, but Cole's September dominance is, at the very least, a good sign for the Pirates.

. . .

All statistics courtesy of FanGraphs, Baseball Savant and Brooks Baseball.

Jake Dal Porto is a writer at Beyond The Box Score and Golden Gate Sports. You can follow him on Twitter @TheJakeMan24.

More from Beyond the Box Score

Follow @BtBScore Follow @SBNationMLB

• Can Mark Reynolds fill the Brewers’ first base void?
• A readily available Bronson Arroyo alternative
• Was Michael Wacha tipping his pitches? A follow up
• How has sabermetrics changed baseball?
• Is this a golden age of catching?