There are three ways to predict breakouts in baseball: Pick a player who broke out the year before and hope no one else noticed; guess blindly and hope to hit on a Jose Bautista by chance; or spend hours poring over every available trend, statistical study, prospect ranking, and dubious report from spring training, and then guess, almost as blindly as you would’ve without any prep.
If you choose the third route, you’ll soon become a connoisseur of mechanical adjustments: streamlined swings, newly compact deliveries, and lengthened or shortened strides, any of which could be the seed that spawns a new superstar. A close cousin of the mechanical adjustment is the New Pitch, the most tangible of all potential tweaks. It registers on radar guns. It’s visible on video. It’s the closest baseball players come to leveling up and unlocking a new ability that lets them beat the next boss. Throwing pitches is the point of the profession, so when a pitcher adds a new type to his collection, it’s worth paying attention. But the flops should also inform our expectations: A complete accounting of what a new pitch is worth considers both those that have worked well and those that went nowhere.
The nice thing about new pitches, from an analytical standpoint, is that they’re (relatively) easy to detect: We don’t have any information on the angle of each hitter’s swing or the height of his hands when he’s in his stance, but we have several seasons of PITCHf/x data that tell us how much each pitch moved and how fast it traveled. One PITCHf/x query can tell us as much about pitch usage as hours of Googling and video comparisons can tell us about mechanics.
For this exercise, we’ll be relying on Pitch Info’s pitch-type classifications, provided by Baseball Prospectus. First, some ground rules: Not every stray pitch that doesn’t fit a pitcher’s usual pattern is a bona fide new pitch type; for example, a lone slider, professional pitch-tagger Harry Pavlidis says, “may have been a really shitty cutter.” What’s more, there are serial experimenters who play around with new pitches but never make a major commitment; below a certain usage threshold, it’s hard to imagine a new pitch having an appreciable impact on a pitcher’s overall stats. (Former reliever Ryan Franklin, for instance, futzed around with a knuckleball, but it was never a regular part of his repertoire.) To filter out the false positives and the hobbyists who never threw their new pitch often enough to worry batters, I looked for pitches that a pitcher threw at least 5 percent of the time (minimum 500 overall pitches) following a season in which he was active but never threw that pitch. The complete PITCHf/x record goes back to 2008 (with partial information from 2007), so I started the search for new pitches in 2009.
I wasn’t a strict constructionist when it came to defining new pitches. For example, as a rookie in 2012, Oakland A’s reliever Sean Doolittle threw 12 pitches classified as sliders in 47.1 regular-season innings, a 1.5 percent rate. In 69 innings the next season, he threw zero sliders. In 2014, though, he remembered that off-speed stuff existed, discovered that his slider had benefited from the time off, and threw it 110 times, an 11.8 percent usage rate. Technically, the pitch was in “barely used” condition, not brand new. But Doolittle’s slider fit the spirit of the search, so I included it, along with any other pitches a player had previously dabbled in less than 5 percent of the time, abandoned for at least one full season, and then brought back with a vengeance. I couldn’t always find a corroborating quote to confirm a pitch was a recent creation, but very often I could, even with the more obscure arms. New pitches are a popular subject.
The winnowing process left me with 121 new-pitch seasons (from 113 pitchers) over the past six seasons, an average of 20 per season. You can scroll through the list of pitchers with new pitches (in descending order of pitch frequency) in the embedded table below. Keep in mind that some pitch types fall on different points of the same speed/movement spectrum, with no clear demarcations between categories, which explains why pitchers don’t always label their offerings the way Pitch Info (or the nearest scout) sees them.
[googleapps domain="docs" dir="spreadsheets/d/1dW17ZOSc3IBRcfCf_AtGYC8q7lUeDfJPQOAgje5g6Ls/pubhtml" query="gid=0&single=true&widget=true&headers=false" width="100%" height="520" /]Using those results, we can break down how often pitches of each type are added, as shown in the following table. The percentages in the third column tell us what proportion of all new offerings the pitches of that type represent.1
| New Pitch | Pitchers | % of New Pitches |
| Cutter | 41 | 33.3 |
| Sinker | 31 | 25.2 |
| Slider | 22 | 17.9 |
| Splitter | 11 | 8.9 |
| Curveball | 10 | 8.1 |
| Changeup | 5 | 4.1 |
| Slow Curve | 2 | 1.6 |
| Four-Seamer | 1 | 0.8 |
1.
Two pitchers added multiple qualifying pitch types in the same season, so the totals in the “Pitchers” column sum to 123.
The four-seamer is the starting point for most pitchers, so it makes sense that Joe Smith, who threw only sinkers until he began to vary his hard stuff in 2010, was the lone player to pick one up. Curveballs and changeups are also part of the standard starting pitcher development kit, so they aren’t common additions among major leaguers. Cutters, sinkers, and sliders, however, are closer to variations on preexisting pitches than they are to stand-alone entities that must be developed from scratch, so they’re easier for pitchers to work out over one winter or one stay at spring training.
What we really want to know, though, is whether adding a new pitch is more than a way for a pitcher to pad the “skills” section of his résumé. Is it also a reliable way to get more guys out?
♦♦♦
Christian Petersen/Getty Images
We can answer that question the same way we sized up the effect of being in the best shape of one’s life: by comparing players’ actual performance to their projected performance heading into the season in which they did something different. BP’s projection system, PECOTA, can generate “retrojections” for previous seasons that tell us what a player would have been expected to do at the time, based on his performance up to that point. We ran retrojections for each pitcher who met the “new pitch” qualifications and compared his projected True Average allowed — a park- and league-adjusted measurement of offensive production, which (like batting average, in normal offensive seasons) is scaled to a .260 league average — to his TAv allowed in real life. If the new-pitch pitchers, as a group, allowed lower True Averages than PECOTA forecasted for them, it could indicate that the new additions made a difference.
The telltale table is only one explanatory paragraph away. The first row is all pitchers who threw their new pitch at least 5 percent of the time. The second row, a subset of the first, is all pitchers who threw their new pitch at least 10 percent of the time. The bottom two rows divide the 5-percenters into two groups — those who added a hard pitch (a four-seamer, sinker, or cutter) and those who added anything else. In this case, negative is good: The further below zero the figure in the “Difference” column, the more the pitcher outperformed his projected TAv allowed.2
| New Pitch Group | Pitchers | IP | Projected TAv | Actual TAv | Difference |
| >=5% | 121 | 11,943 | .2667 | .2612 | -.0055 |
| >=10% | 65 | 6,330 | .2652 | .2582 | -.0070 |
| >=5%, Hard | 72 | 7,595 | .2658 | .2597 | -.0061 |
| >=5%, Soft | 49 | 4,349 | .2680 | .2634 | -.0046 |
2.
Boring methodological detail: I weighted both the projected and actual True Averages of the pitchers in each group by their actual innings-pitched totals. Weighting by the harmonic means or the smaller of the two innings-pitched totals (actual and projected) produced almost indistinguishable results. So, now you know that.
In each sample, the pitchers handled hitters better than PECOTA foresaw. A difference of 5.5 TAv points for the 5-percenters might not sound like much, but over a sample this large, it’s not insignificant. Based on BP’s rule of thumb that each TAv point is worth half a run over 500 plate appearances/batters faced, 5.5 points is worth roughly five runs, or about half a win, over a full season by a starting pitcher. Mastering new pitches isn’t easy, but half a win would be a strong incentive to get good enough with an extra offering to throw it at least one time in 20. Those who threw their new pitches at least one in 10 times enjoyed a slightly larger advantage, which makes sense: Generally speaking, the more a pitcher throws a pitch, the more confident he is in its quality (or the more desperately he needs something new). Even if a new weapon a pitcher is woodshedding doesn’t have out-pitch potential, the uncertainty it plants in the batter’s mind might make it worth throwing, as long as it’s serviceable enough not to get hammered.
One caveat: If PECOTA projected a pitcher to be in the rotation and he ended up in the bullpen (or vice versa), that could skew the result, since relievers tend to allow lower True Averages. On the whole, though, the 5-percenters pitched 372 more innings than they were projected to, and one would expect relievers who are shifting to the rotation to be more likely to expand their arsenals than starters who are moving to a more restricted role. As an additional check, I used the same method to compare PECOTA’s retrojections for the same players in the year before they added their new pitches to their actual results in that previous year. This time, the projected and actual True Averages were nearly identical: .2652 projected, .2648 actual. In other words, PECOTA pegged these pitchers perfectly when they were their same old selves, and then underestimated them by a significant margin once they mixed things up.
Of course, the average improvement doesn’t apply to every pitcher who adds a new pitch: The benefit depends on the pitch’s nastiness and how well it complements the rest of the pitcher’s repertoire. So which pitchers are angling for the new-pitch bump this season?
♦♦♦
Gene J. Puskar/AP
As usual, not every new-pitch spring-training tall tale has come true. Max Scherzer hasn’t used the cutter he threatened to throw (although he did throw more sinkers in his Opening Day start than he did all of last season). Taijuan Walker’s refined cutter is slower and has more vertical movement than the old one, but it’s not distinct enough for Pitch Info to label it a slider. David Price has thrown fewer curveballs than usual despite tinkering with a new curve variant last month. Darren O’Day’s improved changeup, Tommy Hunter’s new splitter, and Archie Bradley’s and Aaron Sanchez’s new cutter/sliders have yet to materialize (although Sanchez has raised his changeup rate). And Justin Verlander has spent the beginning of the season on the disabled list, so he hasn’t had a chance to unveil the mystery pitch he dropped hints about in Florida.
We have seen some significant evolutions in pitch frequency: Shelby Miller, as promised, has accelerated the deployment of the sinker he learned late last year. Similarly, Nathan Eovaldi’s new splitter from late 2014 has carried over to 2015. And as expected, Carlos Martinez, in his capacity as a quasi-starter, has thrown changeups more often than he did last season.
While we’ve been trying to track changes in existing offerings, though, baseball has already brought us a bunch of new business. Many more unfamiliar pitches will pop up throughout the year, some for cameos and some to stay, but the following five new arrivals have made the most frequent appearances in the first half of April. For each new pitch, I’ll include the most comparable pitch of the same type from last season, using Jeff Sullivan’s method for gauging pitch similarity.3 It’s too soon to judge any of these pitches by results alone, but by finding their closest velocity/movement matches, we can get an inkling of how they might fare.4
Jimmy Nelson, Brewers (Curveball)
3.
The Whiff/Swing and GB/BIP scores measure standard deviations above league average, while RV/100 is the pitch’s run value per 100 deliveries, via FanGraphs.
4.
While acknowledging that command, sequencing, deception, and other factors play important roles in determining the effectiveness of a pitch.
[protected-iframe id=”8017e7100b174c17bf05c5420a1876f3-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=GreedyUnawareDassie” width=”530″ height=”300″]
| Closest 2014 Curveball Comp | ||||||
| Pitcher | Velo (mph) | Hor. Mov (in.) | Vert. Mov (in.) | Whiff/Swing | GB/BIP | RV/100 |
| Felix Hernandez | 81.0 | 7.2 | -8.1 | 1.3 | 1.6 | 2.3 |
| Jimmy Nelson | 81.9 | 5.9 | -8.0 | — | — | — |
After watching the Pirates flail at Nelson’s new spike curve, it’s probably not a shock to see it rate well, but a Felix comp is always a pleasant surprise. Last year, Nelson threw almost exclusively four-seamers, sinkers, and sliders — three pitches with small to large platoon splits — to left-handed hitters. Curves have a small reverse-platoon split, and Nelson has a good one, so he should be much better equipped to face hitters from the opposite side.
James Paxton, Mariners (Sinker)
[protected-iframe id=”72c9f5ce8f1e980e7a6cae0e8aeb0590-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=ExhaustedAcclaimedJunco” width=”530″ height=”300″]
| Closest 2014 Sinker Comp | ||||||
| Pitcher | Velo | Hor. Mov | Vert. Mov | Whiff/Swing | GB/BIP | RV/100 |
| Chris Archer | 95.6 | 7.7 | 7.8 | 0.4 | -0.2 | 0.5 |
| James Paxton | 94.7 | 8.0 | 8.1 | — | — | — |
The sinker is a curious choice for Paxton, who gets tons of ground balls with his four-seamer (and his secondary stuff). More high fastballs might be a smart strategy for the big lefty, but his new pitch suggests he’ll continue to live down in the zone, despite pitching in the forgiving fly ball arena of Safeco Field.
Zach McAllister, Indians (Cutter)
[protected-iframe id=”f567839b5c60d8a93bbe65597916b469-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=SelfreliantThoroughGroundbeetle” width=”530″ height=”300″]
| Closest 2014 Cutter Comp | ||||||
| Pitcher | Velo | Hor. Mov | Vert. Mov | Whiff/Swing | GB/BIP | RV/100 |
| Ryan Vogelsong | 88.7 | -0.5 | 7.5 | -0.3 | -0.5 | 0.2 |
| Zach McAllister | 88.6 | -1.2 | 8.0 | — | — | — |
McAllister threw his four-seamer almost three-quarters of the time last season, mixing in some sliders, curves, and changeups. This season he’s dropped the slider (and some of the four-seamers) and replaced those pitches with cutters, another gift from Indians pitching coach Mickey Callaway. As I noted recently, McAllister has also enjoyed one of the largest velocity gains of any pitcher this year, retaining the speed boost he posted in the bullpen in 2014.
Ian Krol, Tigers (Slider)
[protected-iframe id=”af6b4273325a347bbaf0fcb67e8f098d-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=PaleCelebratedChanticleer” width=”530″ height=”300″]
| Closest 2014 Slider Comp | ||||||
| Pitcher | Velo | Hor. Mov | Vert. Mov | Whiff/Swing | GB/BIP | RV/100 |
| Daniel Webb | 85.6 | 1.2 | -2.0 | 2.3 | 2.9 | -1.7 |
| Ian Krol | 88.3 | 0.7 | 4.6 | — | — | — |
Tigers skipper Brad Ausmus was pleased when Krol showed up to spring training with his “cutter-slider” ready to go. The lefty throws his four-seamer in the mid-90s, so he has a speed separation of more than 20 mph between his fastball and his curve. His problem has been a tendency to allow the long ball, which the Tigers hope his new weapon will help curtail. Webb’s slider, the closest pitch comp, misses bats but also misses the strike zone almost half the time, which made it a below-average pitch last season.
Kevin Gausman, Orioles (Curveball)
[protected-iframe id=”e33f4a6ffaad3b46bea781fe548e7859-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=HugePossibleAsiaticlesserfreshwaterclam” width=”530″ height=”300″]
| Closest 2014 Curveball Comp | ||||||
| Pitcher | Velo | Hor. Mov | Vert. Mov | Whiff/Swing | GB/BIP | RV/100 |
| Jake Peavy | 80.5 | 2.4 | -1.9 | 0.51 | -0.23 | 0.36 |
| Kevin Gausman | 81.2 | 1.9 | -2.0 | — | — | — |
Gausman threw 142 sliders last season, and he didn’t get great results: His whiff/swing rate with the pitch was slightly below average, and his ground ball rate with it was two standard deviations below average. The only thing the pitch did really well was allow line drives.
For now, Gausman seems to have scrapped the slider and replaced it with a curve, which he’s thrown a little more often than he threw the slider last year. The new breaking ball sits in virtually the same velocity and vertical-movement range as the old one, but the Orioles consider it a different pitch, a revival of one the righty used to throw in college. “His curveball, potentially, is as good, if not better, than his slider,” Baltimore manager Buck Showalter said, which isn’t the nicest compliment anyone’s ever paid a breaking ball. But Peavy’s was an average pitch, and Gausman has much better stuff to go with it.
♦♦♦
Traditionally, baseball analysts have dismissed April as the sport’s fun-house-mirror month, when it’s safer to close one’s eyes and wait than it is to try to navigate by the distorted small-sample stats that surround us. As the numbers we can crunch get more granular, though, the minimum size of the samples from which we can draw meaning shrinks. Greater awareness of which stats stabilize quickly — and which injuries cluster early in the year — has helped us glean some significance from spring training. Easy access to video has helped us spot changes in appearance that could presage changes in performance. Single pitches, single swings, and, soon, single defensive chances will have much more meaning than they did when the outcome of each play was the only information available. It’s no longer safe for analysts to take April off: The next pitch we see might be someone’s first sinker, and now we know what a new sinker can mean.
Other potential new-pitch sightings, according to Pitch Info data: Christian Friedrich, Rockies (cutter); Nick Hagadone, Indians (cutter); Tyler Matzek, Rockies (sinker); Kendall Graveman, A’s (four-seamer); Tommy Layne, Red Sox (changeup); Jeanmar Gomez, Phillies (curveball); Jose Alvarez, Angels (changeup); Dan Otero, A’s (cutter); Jumbo Diaz, Reds (cutter); Kelvin Herrera, Royals (slider); Adam Ottavino, Rockies (cutter); Greg Holland, Royals (curveball); Chris Tillman, Orioles (slider)
Thanks to Jessie Barbour, Harry Pavlidis, Rob McQuown, Nick Wheatley-Schaller, and Jeff Sullivan for research assistance.