Real Time Analytics

Sample Size Watch, Part Three: Pitchers

Nhat V. Meyer - Bay Area News Group

Jeff Samardzija’s statistically significant rate stats point to a much better pitcher down the stretch.

It might seem a little untimely to discuss sample sizes at the same time as the Mid-Summer’s Classic, but there is a method to this supposed madness. In the first two segments of this three-part series, we examined offensive statistics that stabilize very early, namely strikeout rate and walk rate. The next statistic we might have examined would have been home run rate (HR%), but something as loud and as glamorous as the long ball was unlikely to be overlooked. Everybody knew that Logan Morrison had blossomed into a power hitter, and similarly everybody knew that Xander Bogaerts was in a slugging slump.

Besides, we need to talk about pitchers! Pitchers, however, are flighty beasts, and their statistics take far longer to stabilize than the stats of hitters. Take home runs, for instance. A hitter’s HR% becomes statistically significant after 170 PA. Pitchers, however, require 1,320 batters faced before one can attribute their rates of home runs allowed to skill. The last time one pitcher faced so many hitters in single season was Phil Niekro in 1979. The last non-knuckleball pitcher to do so was Nolan Ryan in 1974.

Why wait until July to talk about pitcher sample sizes? Because, even now, it is still difficult to pinpoint what changes are statistically significant. Nevertheless, several pitcher stats have met their respective thresholds. For all regular starting pitchers, we can confidently turn to their rates of strikeouts (K%), walks (BB%), fly balls (FB%), and ground balls (GB%). For most relievers, we only have strikeouts and in rare instances BB%, FB%, and GB%. With this in mind, here are several arms with some of the most dramatic differences in statistically significant areas.


Zack Greinke (good) and Jeff Samardzija (unlucky)

Image titleChristian Petersen - Getty Images

Pitcher Name
Career K/BB
2016 K/BB
2017 K/BB
Zack Greinke
Jeff Samardzija

Rather than tackle K% and BB% individually for starters, I decided to combine them into one holistic statistic — strikeouts per walk. Before getting to our featured arms, apologies go out to Lance McCullers, the 23-year-old Houston Astros phenom who seems to have learned control (7.1% walk rate improved from 12.8% last year) and harnessed his unbelievable stuff. Even bigger apologies go out to Jimmy Nelson of the Milwaukee Brewers, whose great strides in both his strikeout rate (26.1% from 17.3%) and his walk rate (6.0% from 10.7%) are just several of many reasons why the Brew Crew has a legitimate chance to win the National League Central Division.

Zack Greinke has played as large a role as anybody in the resurgence of the Arizona Diamondbacks this season. After a disappointing D-back debut in 2016, Greinke now resembles a pitcher in his prime. His K/BB rate is the best it has ever been, even with the slowest fastball of his career (90.9 MPH).

Greinke’s return to elite performance could be due to less reliance on that aforementioned fastball or to the benefits of having an elite framing catcher like Jeff Mathis behind the plate. Much credit is due to a noticeably improved slider. As you can see in the chart below (courtesy of Brooks Baseball), Greinke’s slider is getting more whiffs now than at any point in the past five years.

Image title

His slide-piece is a big reason why the righty can maintain the second-best out-of-zone swing rate (38.6%) and third-lowest percentage of pitches in the strike zone (39.0%) among all qualified pitchers. Should these trends continue, Greinke is poised to lead Arizona to October as he looks to put last year’s forgettable season behind him.

Image titleChris Carlson - AP Photo

Meanwhile, in another corner of the NL West, Jeff Samardzija is having a wacky season. He leads the NL in losses (10) and is sitting on a disappointing 4.58 ERA for the last-place San Francisco Giants. At the same time, he leads all starters in K/BB and is posting the best xFIP (3.09) and SIERA (3.26) since his standout season in 2014. He has started more games (18) than walked batters (14) What gives?

In many ways, the Shark has been simply unlucky. His BABIP against (.323), left-on-base percentage (67.1%), and home run to fly ball rate (16.7%) are all worse than average. If these three stats regress to the mean over the next few months, Samardzija is a second-half breakout candidate. However, due to a dismal season from the Giants and his not-so-movable contract (nearly $60 million over the next three seasons), any Samardzija success story will be unlikely to reach the postseason.

Tommy Kahnle and Sam Dyson

Image titlePaul Beaty - AP Photo

Pitcher Name
Career K%
2016 K%
2017 K%
Tommy Kahnle
Sam Dyson

Tommy Kahnle’s story is pretty cool. A Rule 5 draftee designated for assignment by the Colorado Rockies, the 27-year-old righty (28 in August) is now dominating as part of the Chicago White Sox bullpen. Even cooler than his story are the stories being written about him. In April, FanGraphs noted that he had been one of the best pitchers in baseball over the first month of 2017. In June, FanGraphs essentially wrote the same thing. Yet to many Kahnle is still an unknown, despite a K% that is through the roof.

Poor Sam Dyson. A year ago today, Dyson was the closer for a playoff-bound Texas Rangers team. Now he is a member of the Giants, hoping to survive the Texas DFA. His ERA is an unsightly 7.45, his WHIP is 1.97, and honestly none of his numbers look good. That includes a strikeout rate that has declined in each of the past three seasons.

Image titleRon Jenkins - Getty Images

For each reliever, the story is about velocity. Dyson lost a tick on his fastball, sitting at 95.8 MPH after averaging 97.0 last season. As Jeff Sullivan pointed out a month ago, he also cannot reach his max velocity as often. Meanwhile, Kahnle’s velocity has climbed after each season in the Big Leagues. Thanks to his size and extension, his perceived velocity according to StatCast (98.5 MPH) is the third-fastest in the majors. Kahnle looks like he can sustain his high-octane dominance, but there is hope for Dyson, too. In 12.1 innings with San Fran, he has a 2.17 FIP and 12 strikeouts. That sample is far too small to draw meaningful conclusions, but it is a positive step for Dyson out west.


Ivan Nova (he’s lucky but a better pitcher now, too)

Image titleJustin K. Aller - Getty Images

Ground Ball Rate (GB%)
Fly Ball Rate (FB%)

Nova was another candidate for the K/BB section of the article because, like Samardzija, he has more starts (18) than walks (15). In the past calendar year, Nova is sixth in ERA among qualified pitchers. His reclamation is counted as yet another win for Pittsburgh Pirates pitching coach Ray Searage. What’s interesting, though, is how Searage and the Pirates changed him.

Over the past year, Nova’s GB% and FB% have flipped a bit. To be fair, Nova is still a ground ball pitcher, owning a GB/FB rate of 1.63 this season versus the league average of 1.25. Whereas Greinke has started to throw his slider more often, Nova is relying on his primary offspeed pitch — his curveball — at a lower rate (20.3%) than at any point in his eight-year career. Going to his fastball more often has led to a higher number of fly balls, but that has not been bad news for Ivan. Here is a depiction of his fly balls per balls in play off Nova’s pitches over the last two years:

Image title
And here is the isolated power (ISO) against each of Nova’s offerings over that same time frame:

Image title
The change of scenery helps: Yankee Stadium is third in park factor for homers, whereas PNC Park is 25th. The confidence to pitch in the strike zone has resulted in the second-highest Zone% (50.4%) in his career. However, there are warning signs around Nova. Unlike Samardzija, his BABIP (.267) is well below the league average and his LOB% (79.1%) is far above. His 4.17 xFIP, while still solid, is nearly a run higher than his ERA (3.21). Whether he will regress or not, Nova the Pirate is clearly a different pitcher since he first came to Pittsburgh.


Rick Porcello (unlucky but worse than last year) and Justin Verlander (too many walks!)

Image titleDan Hamilton - USA Today

Pitcher Name
Career OBP Against
2016 OBP Against
2017 OBP Against
Rick Porcello
Justin Verlander

I am cheating here a bit. Technically, OBP does not stabilize until 540 batters faced. Porcello has faced the most hitters in baseball with 525 BF. Verlander is far behind at 467 BF. But these two starters are so interesting because they hardly resemble the Cy Young winner and runner-up, respectively, that they were in 2017.

Porcello led the American League in wins (22) and K/BB (5.91) last season. Today, he leads the AL in losses (11), hits allowed (147), and earned runs (63). As was said at the time of his breakout season, there would be some expected regression for the Boston Red Sox sinkerballer. His FIP (4.07) indicates he’s pitched better than his ERA (4.75), but he also has the lowest GB% of his career (37.8%, far below his career number of 49.2% and likely due to his lowest rate of sinker usage). He has not lost any velocity and his peripherals are not dramatically different, so expect Porcello to pitch fine down the stretch and lower his OBP against by a fair margin. It will take a lot for him to ever match his Cy Young season, but the Red Sox already have an ace on their staff in Chris Sale.

Image titleRick Osentoski - USA TODAY Sports

Porcello has been bugged by base hits, whereas Verlander has been victimized by walks. Last year, Verlander walked 57 hitters in 227.2 innings. Halfway through 2017, Verlander has 51 walks – tops in the American League – in just 104.2 innings. If walks were the only problem, maybe his case would seem more salvageable. However, the once-ace of the Detroit Tigers is getting fewer swings and misses on all of his pitches and has been hit harder than at any point in his career.  The best hope for the 34-year-old righty is that this .345 OBP is not representative of his true talent.  To reach that stabilization mark of 540 PA, Verlander would likely need three more starts. It would be foolish to count out this Cy Young runner-up, though his statistics thus far do not paint a pretty picture.


This series began as a way to respond to the cries of “It’s still early” at the outset of the season. Together, we learned that there are statistics that tell stories even in April. Some of those tales continue to this day — Aaron Judge has exhibited improved control of his strikeout rate, and his power has played up accordingly. It goes to show how small trends beneath the surface can have huge and sometimes legendary impacts.

While this is the last entry for the 2017 season, hunting for meaningful statistical data never really ends. Larger samples just mean greater clarity into a player’s true talent level. And small samples of certain rate stats are still useful when examining a player’s rolling performance. Just wait until call-ups after the July trade deadline or roster expansions in September, when stat-watching becomes just as fun as in April.

All statistics accurate through July 11th, 2017 and were obtained at Baseball-Reference, Baseball Savant, Brooks Baseball, and FanGraphs, respectively.

Edited by Joe Sparacio.

How many wins did Justin Verlander have in his 2011 MVP/Cy Young season?
Created 7/11/17
  1. 19
  2. 20
  3. 22
  4. 24

Join the conversation! 2 comments


What do you think?

Please log in or register to comment!

Can't get enough SQ?

Sign up for our weekly newsletter here!