How Chad Kuhl Bases The Curveball Off The Slider

Chad Kuhl had an up and down season, seeing an increase in strikeouts, but also walks, in his second season.  His adjusted ERA and FIP were right around league average, just like in 2016.  However, he added a curveball at the end of May, and he had solid results with it.

Chad Kuhl finished 2017 with a 4.35 ERA and 4.24 FIP, when adjusting for the park and league those are two percent worse (102 ERA-) than league average and league average (100 FIP-) respectively.  I wrote earlier about Kuhl’s progression as a pitcher, and it featured a similar graph of his strikeout rate:

In late May and early June, Kuhl saw his strikeout rate really increase, with some small decreases, but it trended up as the season kept going along.  Part of that reason is that Kuhl added a curveball, a pitch to help against left-handed hitters.  Alex Stumpf of The Point of Pittsburgh wrote an excellent piece about how Kuhl’s Curveball Is More Than A Pitch, It’s An Idea earlier this offseason.  This part of Stumpf’s article really got me thinking:

“The most obvious difference was brushing off a curveball that the Pirates, as he described it, “put on the back burner” when he entered the farm system. To his surprise, throwing a hook is like riding a bike, and he increased its usage every month this season.

“It’s nice to have something that can change eye level, can change speeds, as simple as it may be,” Kuhl said in an interview with TPOP. “It’s another weapon, and I think that enhances everything else that I throw.”

All of a sudden, lefties were no longer a problem, or at the very least less of a problem. Southpaws recorded a .342 wOBA in those last four months, which resulted in a 3.75 ERA. And it was not just the curveball that was throwing lefties off. All of his pitches were more effective against them. Like he said, the pitch “enhances everything else.”

Kuhl increased his curveball percentage each year, you can see that both in Alex’s piece and my piece, and it really added to him improving.  The thing that caught my eye is how Kuhl said that “It’s another weapon, and I think that enhances everything else that I throw.”  Kuhl believes his curve enhances his other pitches, but what if the other pitches also enhance his curve?  What if instead of just using his slider as an out pitch, he also uses the slider to help set up the curveball as an out pitch?

Eno Sarris wrote and talked with Sonny Gray about pitch grips and Gray’s thoughts on how grips don’t matter, which gave me the thought that perhaps Kuhl is doing something similar to produce two out pitches, basing the new breaking ball on the one that he has used since entering the Pirates system.  Kuhl’s two pitches are distinct, but the thought he can use one breaking pitch to set up another intrigued me.  I decided to take a look at all the pitch pairs that Kuhl threw in his career, but I really only focused on the curveball, and the slider/curveball sequence, and also looked at the pitch pairs since adding the curve on May 31st.

Methodology

Baseball Savant – which is part of Major League Baseball Advanced Media (MLBAM) – has pitch-by-pitch data going back to 2008, and since Statcast has been installed in 2015, the site also contains data related to that.  Once downloading the pitch-by-pitch data for every game by Kuhl in a CSV file, which is rather simple using the search feature, I was able to string together pitch sequences.  It is important to note that Baseball Prospectus has pitch pairs on their site, and these results will be similar, not a perfect match.  Baseball Prospectus has pitches reclassified, where as Savant does not.  This is an important difference, as the data on Baseball Prospectus is more accurate when it comes to the pitch classification, but this method will still give us a broad look of how Kuhl sets up his curveball and how it is used.  One of the reasons I like this method of using Savant over Baseball Prospectus is because it allows us to see how a pitcher pairs his pitches after adding or subtracting a pitch from his arsenal around a certain date and forward, whereas Baseball Prospectus is just the season data.  This is a repeatable process for Kuhl, and any pitcher, along with any pitch, and this is the process in which I went about it:

  1. Download pitch data from Baseball Savant into a CSV and open in excel or similar spreadsheet application.
  2. Sort the date by oldest to newest, sort at_bat_number least to greatest, and sort pitch_number least to greatest.
  3. Create a new column directly after pitch_type and call it something such as “pitch_pairs” or “sequence.”  Use the ampersand (&) to combine the pitch_type column to create pitch pairs.  For instance, if I want the sequence to be in column B and the pitch_type is in column A, I would click into cell B2 and type =A1&”.”&A2.  The result of this would create A1.A2 in cell B2, but you can use whatever character in between the quotations to separate the two values in each cell.  Using a concat function would work as well, this would look like =concat(A1,”.”,A2).  Either method gives you the pitch pairs, and once the pitch pair is obtained copy the cell formula all the way down the spreadsheet.
  4. Set up a new column after the at_bat_number column named something like “new_batter.”  Pitch pairs really only matter for the new batter, a first pitch out a fastball followed by a first pitch out on fastball isn’t really that important for this purpose.  Use an IF function to set this up.  If the at_bat_number is in column C, in column D your IF function would look something like =IF(A1=A2,”no”,”yes”).  The “no” would indicate that this is not a new batter and the “yes ” would indicate that it is.  Copy the formula all the way down the sheet.
  5. Filter the column in which you used the IF function, this would be the column you just created called “new_batter,” and filter it such that only the “yes” would appear.  This would signal the sign of a new batter, and we no longer need the pitch pairs for it.  Highlight the cells, important you highlight the cells and not columns because you only want to eliminate the new batters, and delete the “pitch_pairs” or “sequence” data for those new at bats.
  6. Remove the filter for the “new_batter” column and then you will have the pitch pair data.  You can compare the data to Baseball Prospectus’ pitch pair data, just remember that it will not match perfectly due to 1). reclassifying the pitch types and 2). Baseball Prospectus likely uses a different method.  This method of using Savant’s data and excel is simple and anybody can do this just by using a spreadsheet, and the results for the pitch pairs are very similar.

The other nice thing with the Savant data is how any calculation on the pitch can be done, including their usage of a pitch, the counts they use them in, and because of ability to filter by date, the frequency and usage since a certain date (for instance since Chad Kuhl began to throw the hook).  Other data such as horizontal and vertical movement, where the ball was caught, and different events and descriptions can all be calculated through a pivot table and filtering.

Adding the curveball

After going through the method above, I used Tableau to visualize the data and noticed some things.  The first thing that I did was look at where Kuhl got his strikeouts on the curveball and plotted the plate x and plate z coordinates with a rough approximation of the strike zone.  This is really no different from Savant’s search feature and using one of their graph options, but this is more interactive and can change the pitch type and receive results faster using this Tableau instead of having to do a search on each pitch type and result type.  This should work on a mobile device or tablet, but would be much better in landscape mode.
These are just the curveballs that resulted in strikeouts, you can adjust the Tableau to get all his curveballs with any event or description, and you can do this for any pitch he’s thrown, and this is all from the catcher’s point of view.

All but one of Kuhl’s swinging strikeouts occurred down and away to right-handed hitters or down and in to left-handed hitters.  The looking strikeouts on the curve were all very hittable pitches, but the hitter was likely not expecting them, as they occurred middle and in for a right-handed hitter and middle and away for a left-handed hitter.  Kuhl had 15 strikeouts on the curve (12 swinging), and he only threw it 163 times, meaning that the result of a curveball thrown by Kuhl was a strikeout 9.2 percent of the time.  Using the Baseball Savant search feature and a lookup function in excel, Chad Kuhl ranked 38th among the 123 pitchers to throw 150+ curveballs since May 31st in strikeouts on curveballs/curveballs thrown.  May 31st was the first game in which Kuhl actually started to throw the curve, and that’s a good starting point for this exercise.  There is a thing we can learn from this, Kuhl looks to throw the curve ahead of the count when looking for a strikeout, in fact, 40.49 percent of Kuhl’s curves came when there were two strikes on the hitter.

It would appear as if Kuhl would like to use his curve as his out pitch, but he also has a slider that he likes to throw as well.  Starting with May 31, this is a breakdown of the counts in which Kuhl threw his slider and his curve, with the columns being how many balls there were and the rows being strikes.  This is a percent of the pitch thrown, so, for instance, the 21.47 percent in a 0-0 count means that 21.47 percent of Kuhl’s curveballs were thrown in a 0-0 count.

Curve0123
021.47%3.68%0.61%0.00%
123.31%8.59%1.84%0.00%
29.20%18.40%12.27%0.61%
Slider0123
022.39%7.38%2.54%0.00%
112.47%7.38%6.11%0.51%
28.40%15.27%13.49%4.07%

Kuhl likes to go to hook most often as a get me over curve, ahead 0-1, and ahead 1-2.  His slider on the other hand, he used primarily at 0-0, up 1-2, and tied 1-2.  In total, 211 of Kuhl’s 416 pitches (51 percent) in an 0-2, 1-2, or 2-2 count came via the curveball or slider.  Instead of trying to use a traditional three pitch mix (fastball, changeup, and either a hook or slider), Kuhl has been going with a four pitch mix, utilizing the curveball and slider as his out pitches.

Setting up the curveball from the slider

The way to look at how he sets up his curveball, I took a look at the pitch pairs that can be created using the previously discussed method.  On the left side is the pitch setting up the curve and on the right is the pitch setting up the slider.  CH.CU means that Kuhl threw the curveball after a changeup only once (adding in the knuckle curve “KC” means he threw the change 11 times before the curve).  These pitch pairs are from May 31st and on since that is when Kuhl added the hook to his arsenal of pitches.  Pitch abbreviations can be read here.

Null represents no pitch that was paired, so all 35 nulls for the curve and 88 nulls for the slider occurred on a 0-0 count.  The most interesting thing that is seen in the pitch pairs is the 28 times that Kuhl followed up a slider with a curveball.  Among the 163 curves that Kuhl threw, 17 percent came after the slider, and of the 128 pitches Kuhl threw directly before a curve to a batter, 22 percent came after the slider.  Six of Kuhl’s 15 strikeouts on the curve came the pitch after the slider, and six of the 18 swinging strikes on Kuhl’s hook came after a slider:

Basing the hook off a slider was interesting to me, so I looked at Baseball Prospectus’ pitch pair and pitch tunneling data to get a feel for how many other pitchers have thrown 25+ curveballs directly after a slider (Baseball Prospectus has it occurring 27 times for Kuhl, one of those important notes I mentioned earlier).  Among the 43 pitchers to do so, Kuhl had the sixth lowest release difference (2.1 inches), releasing the curve and slider at similar arm slots.

This is neither good nor bad, Justin Verlander has the fourth lowest release difference whereas Clayton Kershaw has the fourth highest.  It’s not just his release point that is consistent though, when Kuhl goes slider to curveball, his tunnel differential – Baseball Prospectus defines this as “This statistic tells you how far apart two pitches are at the Tunnel Point—the point during their flight when the hitter must make a decision about whether to swing or not (roughly 175 milliseconds before contact)– is the fourth lowest, the two pitches being separated by 8.9 inches on average.

So far, Kuhl has released his slider and then the curveball at similar points and the pitches themselves are closer together than the league average at the time in which a batter makes his decision on if he should swing.  When you lower the amount of times a pitcher has gone slider to curve, Kuhl still has the eighth lowest release point difference and sixth lowest tunnel differential among the 59 pitchers.

Based on what Eno showed about Gray, I took a look at if Kuhl has any real overlap between his slider and curveball, and if there’s really any third breaking ball.

 

There was essentially no overlap between the curveball and slider, as they feature different horizontal and vertical movement.  Kuhl just throws two different breaking balls, a slider and curve, with no tweener pitch.  These breaking balls are two pitches that help play off each other, especially slider to curve, as seen with the Baseball Prospectus pitch pair and tunneling data.

Final thoughts

Chad Kuhl has had success with his slider and curveball individually.  Pitch values, which is a linear weight, is a way to see how well a pitch has actually been, and Fangraphs describes why to use pitch values:

“Pitch values allow you to see how “successful” each pitch has been over the course of a season. You may know that a hitter has a .350 wOBA or that a pitcher has a 3.10 FIP, but there are more fine grained details to consider. Has a particular pitch been effective for a pitcher? Do hitters struggle to produce against sliders? When the hitter puts a changeup in play, is it usually an out or a home run?

Pitch Values are an accounting method for attaching run values to each specific pitch rather than each specific plate appearance. There are reasons to be cautious, discussed below, but if you want a retrospective look at which pitches were crushed and which were effective, pitch values will provide you with some approximate totals. If you see a wFB/C of 1.50, you can generally say that hitter was successful against fastballs that year. It’s not always that simple, but that’s the basic idea.”

Since Chad Kuhl finished 4.2 innings short of qualifying for the ERA title, I adjusted the Fangraphs leaderboard to a minimum of 130 innings.  The goal was to be able to get those pitchers who were called up to avoid being super two eligible for arbitration, and lowering the minimum innings to 130 gave near 100 results (120 innings had 115 results, 130 innings had 105 results,140 innings had 90 results).  Kuhl, among the 105 pitchers, had the 25th highest pitch value with the curveball based on the pitch info data with 4.7 runs.  His slider was even better, ranking 16th in pitch value with 13.6 runs.

Pitch values are not predictive and can be misclassified, therefore an awareness of the sample size of each pitch is needed, Fangraphs even offers that disclaimer.  In the 2016 season, Kuhl was +5.2 runs on his slider.  It’s a good offering by Kuhl, and it appears that the curveball has the potential of being a good one as well.  His sinker, fastball, and changeup, on the other hand, have not been good offerings.  His 97 mile per hour fourseam fastball might have to be his pitch going forward, given his sinker has had run values of -8.1 and -4.1 these last two seasons.

Chad Kuhl’s curveball and slider look like they have the potential to be solid offerings and out pitches going forward, and the way he plays the curve off the slider is rather interesting.  Given the proximity in the release, how close the two pitches are when a hitter has to make a decision, and the six mile per hour difference in velocity – Kuhl’s curve averages 82.3 miles per hour and his slider has averaged 88.2 mph in his career using Savant’s data – can allow him to use the slider to help set up the curve.

Perhaps Kuhl’s curve is enhancing his other pitchers, but it can also be that the slider is helping to enhance the curve.  Similar to Gray, Kuhl throws two different breaking balls, though there really is no overlap between the pitches.  The two breaking balls is a setup that might work out for Kuhl, considering his lack of success in the change.  With more comfort in the curve going forward, I’m interested to see if Kuhl keeps pairing the slider followed by the curve in 2018, and if he keeps getting positive results from both pitches.

*Data from Baseball Savant, Baseball Prospectus, and Fangraphs