This is the 5th article in a series of articles about the beautiful creation known as Statcast. I encourage you to check out the previous articles on four-seam fastballs, cut fastballs, curveballs and sliders. Statcast is a resource provided by MLB.com via the website baseballsavant.mlb.com. There seems to be an endless amount of data here to mine…and it’s all at our fingertips. The question is: how can we use this data to help us fantasy baseball addicts in our baseball player analysis for fantasy baseball purposes? In reality, this Statcast data is merely multiple pieces to a larger puzzle when analyzing each individual baseball player. It is new and exciting data that deserves analysis, but we shouldn’t lose sight of all the other resources on the Internet at our disposal.
Currently I want to focus on starting pitcher analysis with Statcast data. However, just not blanket analysis. Drilled down analysis looking at a single pitch. Today I want to look specifically at knuckle curves, sometimes also referred to as spike curves.
Now we’re going to get into some statistics that are courtesy of baseballsavant.mlb.com. I want the readers to know for a starting pitcher to enter the data that I pulled for analysis they had to have a minimum of 100 total pitched thrown in that season (although I dropped that down to 10 in the current season).
This pitch is an interesting creation, because while curveballs are known to typically spin at high rates…well…knuckleballs don’t.
The first lesson that we learned in the previous articles, in regards to four-seam fastballs, cut fastballs, curveballs and sliders, is that, even if you don’t have elite level spin rate on it as a pitcher, that doesn’t mean you are doomed to it being an ineffective pitch. That lesson applies to knuckle curves as well.
Usually at this point in the article I say “the second lesson” is that even if you do have elite level spin rate on the pitch (four-seam fastballs, cut fastballs and curveballs) that doesn’t mean you are guaranteed for it to be an effective pitch. In the previous article about sliders this statement held true…it isn’t a blanket guarantee…but when over 88% of the starting pitcher performances that had elite average spin rate (2 of 18 performances in that two and a half season data set) on their sliders have batting averages allowed of under .227…well, lets just say this: there is a high likelihood of effectiveness with elite level average spin rate sliders. Of course, we aren’t talking about sliders in this article today. It is about the knuckle curve. I’ve looked at the average spin rate numbers from qualified performances from the data sets of 2015 (27 performances), 2016 (24 performances) and so far in 2017 (23 performances) and I haven’t seen anything emerge with average spin rate…high or low…being something that indicates a higher likelihood of success with the knuckle curve.
However, never fear, because there are many other factors that lead to whether a pitch thrown will be successful. Some of them are (in no particular order): velocity, pitch placement, talent level of defensive player ball is hit closest too, where the ball it hit, how hard it is hit, etc.
Follow The Elite Whiffs%:
Well, it may not highlight a lot of performances, but starting pitchers with whiffs percentages (whiffs% = whiffs divided by swings) of 40% or higher have given up only low batting average against numbers since 2015. Here is the group of performances with a knuckle curve that produces a 40% or higher whiffs% from 2015 to today from the data sets I collected:
In 2015 those pitchers were:
James Paxton: 1799 average spin rate, 159 knuckle curves thrown, .164 batting average, .381 BABIP, 41% whiffs%
James Shields: 2445 average spin rate, 626 knuckle curves thrown, .191 batting average, .281 BABIP, 40.7% whiffs%
In 2016 those pitchers were:
Lance McCullers: 2914 average spin rate, 664 knuckle curves thrown, .140 batting average, .263 BABIP, 42.6% whiffs%
Alex Wood: 1916 average spin rate, 287 knuckle curves thrown, .182 batting average, .333 BABIP, 41% whiffs%
Gerrit Cole: 2626 average spin rate, 189 knuckle curves thrown, .188 batting average, .346 BABIP, 40% whiffs%
In 2017 those pitchers are:
Nate Karns: 2307 average spin rate, 269 knuckle curves thrown, .169 batting average, .303 BABIP, 48.3% whiffs%
Trevor Cahill: 2924 average spin rate, 264 knuckle curves thrown, .127 batting average, .233 BABIP, 45.6% whiffs%
James Paxton: 2038 average spin rate, 376 knuckle curves thrown, .148 batting average, .302 BABIP, 40% whiffs%
Analysis: Eight performances isn’t a large group. However, this does show, so far anyways, that 40% or higher whiffs% starting pitchers in the data set produce low batting average against numbers. Of course, logic suggests that a high whiffs% will often lead to a low batting average against. However, at least from these combined data sets, 100% of the time that produces a batting average at or under .191.
When we look at the starting pitchers in the combined data sets that have a 38% to 39.99% whiffs% we see batting averages of .246, .262, .244, .237, .219 and .216
Before we get to the conclusion I want to briefly look at the other performances of the eight starting pitchers listed above in this article (provided they made the combined data sets).
Lance McCullers (2015): 2686 average spin rate, 763 knuckle curves thrown, .157 batting average, .252 BABIP, 34.1% whiffs%
Alex Wood (2015): 1759 average spin rate, 641 knuckle curves thrown, .220 batting average, .345 BABIP, 36.7% whiffs%
Gerrit Cole (2015): 2481 average spin rate, 225 knuckle curves thrown, .270 batting average, .346 BABIP, 34.2% whiffs%
Nate Karns (2015): 2308 average spin rate, 704 knuckle curves thrown, .173 batting average, .239 BABIP, 30.2% whiffs%
James Paxton (2016): 2030 average spin rate, 265 knuckle curves thrown, .219 batting average, .410 BABIP, 38.7% whiffs%
James Shields (2016): 2531 average spin rate, 512 knuckle curves thrown, .237 batting average, .303 BABIP, 38.9% whiffs%
Nate Karns (2016): 2372 average spin rate, 610 knuckle curves thrown, .228 batting average, .329 BABIP, 29.4% whiffs%
James Shields (2017): 2544 average spin rate, 186 knuckle curves thrown, .218 batting average, .308 BABIP, 35.7% whiffs%
Lance McCullers (2017): 2878 average spin rate, 839 knuckle curves thrown, .207 batting average, .331 BABIP, 36.2% whiffs%
Alex Wood (2017): 2098 average spin rate, 370 knuckle curves thrown, .216 batting average, .338 BABIP, 38.3% whiffs%
Gerrit Cole (2017): 2695 average spin rate, 243 knuckle curves thrown, .250 batting average, .294 BABIP, 24.4% whiffs%
Analysis: These are seasons before and after the noted performances in the list earlier on in the article. Here we have 11 performances, and of those 11 performances only 2 had a batting average above .237. The group, overall, performed very well despite not producing even one 40% or greater whiffs% performance in the season prior to, or after the 40% or higher whiffs% performance that got them noticed. Put as much or as little stock in this as you desire. This entire article on the knuckle curve is dealing with a relatively small sample size (in comparison to the other pitches), but even despite that fact, this is an interesting nugget to chew on.
The conclusion for me on knuckle curves, as far as an easy to sort pattern, is this: Look for the starting pitchers with the knuckle curve that can generate 40% or higher whiffs% and place some faith in the small sample size that in the next season they will be similarly effective with the pitch in regards to batting average against, even if the whiffs% ends up being lower.