Fear Factor: How Pitch Location Helps Reveal Batter Breakouts

There was nothing auspicious about the box score for Kris Bryant’s big-league debut. Bryant, the former second-overall draft pick and reigning best prospect in baseball, broke in on April 17 after a seven-game, .679-slugging-percentage service-time sequestration at Triple-A Iowa. And with the eyes of every non-blacked-out MLB.TV subscriber upon him, the 23-year-old went 0-for-4 with three strikeouts against Padres starter James Shields and reliever Dale Thayer in Chicago’s 5-4 loss, spawning hundreds of almost identical Twitter jokes about the phenom being a bust.

In the same game, extremely un-hyped Cubs second baseman Jonathan Herrera went 1-for-2 with a single and a sacrifice bunt. By Win Probability Added, Herrera’s performance helped the Cubs almost exactly as much as Bryant’s hurt them, increasing their odds of winning by 18.5 percent while Bryant sapped them by 18.6 percent.

But it was only a matter of time until Bryant outhit Herrera. Not much time, as it turns out: Less than two weeks later, Bryant is batting .318/.455/.409 to Herrera’s .235/.257/.294. Let’s assume for a second, though, that we hadn’t known Bryant’s backstory or been aware that Herrera was a 30-year-old middle infielder with a career 69 OPS+. How might we have been able to tell that Bryant was a better player after one game, without even watching?

Herrera is 5-foot-9 and Bryant is 6-5, which is one sign: The correlation between height and Isolated Power (ISO, or SLG-AVG) last season was .24, which means that as height increases, power tends to increase. Bryant batted cleanup and Herrera hit ninth (behind Cubs starter Jason Hammel), another helpful hint.¹. And Herrera laid down a sac bunt, something Bryant hasn’t done since his freshman year at the University of San Diego (and probably won’t do during his MLB career). Put those clues together, and it would’ve been immediately obvious that Bryant had the potential to be the better hitter. But would we have been able to arrive at that conclusion if all we’d known was where the two players were pitched?

^1.

Although batting cleanup in one’s big-league debut isn’t always a harbinger of success.

Here are the locations of the pitches Bryant saw in his first game:

And here are the locations of the pitches Herrera saw the same day:

Six of 14 pitches to Herrera landed in the plot’s nine central squares, including two pitches directly down the middle. Only two of 17 pitches to Bryant fell inside the same squares, with none directly down the middle.

We can pin down that difference in location more precisely. Baseball Prospectus assigns a “called strike probability” to every pitch thrown, based on location, count, year, pitch type, and batter/pitcher handedness. The higher the called strike probability, the less likely a pitch is to be a ball if the batter decides to take. The three pitches in Bryant’s first plate appearance (all of which were swinging strikes) had called strike probabilities of 71 percent, 7 percent, and zero percent, respectively.

[protected-iframe id=”3b9b897738d4960bcadb7481ecc58b5b-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=CostlyPlasticArchaeocete” width=”530″ height=”294″]

The average MLB pitch last season had a called strike probability of 47.5 percent. The called strike probability of the pitches Bryant saw on the day of his debut was less than half that: 22.4 percent. (Herrera’s average was 51.5 percent.) The numbers alone tell the same story anyone watching would’ve seen: Padres pitchers challenged Herrera and stayed away from the heart of the plate against Bryant. That pattern was a sign of respect that labeled Bryant as a hitter who won’t have many more 0-for-4s ahead of him.

Pitchers don’t pound the zone against a player unless some combination of past performance, advance scouting, and pitcher’s intuition tells them they can get away with it, so a hitter’s average called strike probability alone tells us something about how big a threat he is. The graph below plots ISO against called strike probability for every batter-season in the PITCHf/x era with at least 1,500 pitches seen.

The trend is easy to see: The higher ISO is, the lower called strike probability tends to be, and vice versa. The correlation between ISO and CSP is -0.49, which puts them halfway between a nonexistent relationship and a perfectly intertwined one in which the two stats rise and sink like opposite ends of a seesaw. The correlation between CSP and True Average (TAv), BP’s all-inclusive offensive rate stat, is a slightly weaker but still significant -0.32. Of course, a hitter’s willingness to swing at pitches outside the strike zone plays some part in how far away pitchers work. The nine batter-seasons with the lowest average called strike probabilities all belong to one of three players: Pablo Sandoval, Vladimir Guerrero, and Alfonso Soriano, all of whom are known for a lack of plate discipline as well as a profusion of power. On the whole, though, the less pitchers fear a hitter, the more probable strikes he sees.

The table below lists 2014 leaders and trailers in called strike probability, with a minimum of 1,000 pitches seen. There are good and bad seasons on both sides of the list, but the low-CSP group is much more formidable.

Last year, Rob Arthur — then of Baseball Prospectus, now of FiveThirtyEight — made one of those analytical breakthroughs that’s so sensible it seems deceptively obvious in retrospect. Arthur realized that the way a pitcher approaches a hitter isn’t only a byproduct of past performance; it’s also a potential indicator of future performance. Arthur created a stat called “zone distance” — the average distance of the pitches a hitter sees, relative to the center of his strike zone — to measure the hitter’s capacity to make pitchers stay away. By searching for players whose zone distances decreased or increased sharply over the course of a season, he found that he could identify hitters who were good candidates to surpass or fall short of their projections in the following season. The pitchers, in some cases, were ahead of the projection systems: They could seemingly tell very quickly when a hitter’s ability had changed, whether because of a mechanical alteration, an injury (or recovery from one), or some other adjustment.

With the help of called strike probability, we can conduct a similar search to find hitters whose early-season pitch pattern should raise or lower our expectations. For a typical batter, CSP stabilizes, becoming an accurate representation of a hitter’s true talent, in roughly 300 pitches, or 75-80 plate appearances.² We’ve come to the point in the season when hitters are crossing that threshold, so we can look for the players whose called strike probabilities have changed the most in 2015.³

^2.

In more technical terms, 300 pitches is around when the correlation between the small-sample CSP and the full-season CSP reaches 0.7.

^3.

Changes in fastball percentage can say something, too, but it’s too soon for those rates to stabilize — though you’re welcome to look at a too-early list of hitters who’ve seen more or fewer hard pitches this season.

As a proof of concept, let’s review how well this method would have worked last year. The table below lists the hitters whose called strike probabilities changed the most over the first 300 pitches of 2014, relative to 2013 (min. 1,500 pitches in 2013). The left side shows the hitters whose called strike probabilities fell most sharply — theoretically, the ones who were suddenly striking more fear into pitchers than they had the year before. The right side shows those who were making pitchers more daring. The numbers listed in the “Proj.” column are projected True Averages from PECOTA, BP’s projection system, and those listed under “Act.” are actual results. If our approach is sound, we’d expect to see the hitters with falling CSPs outperform their projections.

As expected, seven of the 10 players on the left beat their PECOTA projections, while eight of the 10 on the right failed to match their projections. That’s 15 out of 20 correct calls, a 75 percent success rate. The lists weren’t without some big misses: Evidently, pitchers believed in Ibanez’s 29-homer 2013, which tied Ted Williams for most home runs by a 41-year-old, but the lefty was washed up at 42. The method also significantly underestimated A.J. Pollock and Anthony Rendon, who enjoyed breakouts that neither this system nor the pitchers who bore the brunt of their bats saw coming. But most of the news is encouraging.

Pitchers anticipated a bounce-back from Alcides Escobar and strong seasons from Lorenzo Cain, Adam Lind, Chris Carter, and others. They also bought into the Seth Smith renaissance: Coincidentally or not, the 32-year-old Smith has posted a 137 wRC+ in 621 PA since undergoing a second LASIK procedure in August 2013, compared to his 106 wRC+ in a larger, pre-second-surgery sample in his prime. If the touchup gave him the power to stare into pitchers’ souls, those pitchers can sense it.

Pitchers also acted less cautious around a number of hitters who had down years by their usual standards, due to injury (David Wright, Carlos Beltran) or advancing age (Miguel Cabrera), or who threatened to wash out of the league (Dan Uggla, J.P. Arencibia). They also lost their wariness of Domonic Brown, who mysteriously cratered at age 26 after his breakout 2013. Using CSP as an early-warning system could have saved Allen Craig’s fantasy owners a lot of outs. And had I posted the next five CSP risers, the list would’ve included four more notable letdowns: Chris Johnson, Andrelton Simmons, Mark Reynolds, and Brandon Phillips.

The method isn’t that successful every year, but over the past three seasons, it has a 66 percent success rate in picking over/unders, with CSP fallers overperforming by an average of seven TAv points and CSP risers underperforming by an average of nine points. Pitches fled from Chris Davis at the beginning of 2013 and from Jose Bautista throughout 2010 and in early 2011, presaging their big breakouts. Matt Kemp, who made the top 10 CSP fallers heading into 2012, appeared on the top 10 CSP risers in his lost 2013. CSP promised career years for Carl Crawford and Colby Rasmus in 2010 and heralded the rapid arrival of the end of the line for Kevin Youkilis and Orlando Hudson in 2012.

So who’s next? Here are this year’s called strike probability fallers and risers with a minimum of 200 pitches through Tuesday’s games:

The Angels have a heavy presence on the left-hand side of the list. I covered the continued re-zoning of Mike Trout last week: When Trout took over baseball in 2012, pitchers were slow to scare. Even at the end of that should-have-been-MVP season, his CSP was 48.7 percent, 27th-lowest in the league. Today, it’s 37.7, the second-lowest in the majors after Evan Gattis, who’ll happily hit balls nowhere near him. Trout is still raking, but pitchers are no longer making it easy on him. The surprising thing is how long it took them to catch on.

[protected-iframe id=”64781cd1e24463ae6d4428a844605282-60203239-35703816″ info=”http://wheatleyschaller.com/dev/video_embed.php?id=GroundedSoggyAnkole” width=”525″ height=”548″]

Meanwhile, Albert Pujols hasn’t hit well yet, but pitchers have regained some of their former respect for him after he stayed healthy for a full season last year and came close to hitting 30 homers in a tough park for power, which made him look a lot better than the 2013 version. The third Angel on the list, Kole Calhoun, was underrated by everyone before his four-win 2014. He’s off to an even better start this season (.309/.365/.500), and pitchers aren’t skeptical.

Reds shortstop Zack Cozart might be the most intriguing player on the list. The perennial light hitter is off to a .282/.313/.513 start that would be easy to dismiss as unsustainable if not for the fact that pitchers seem to be buying it. Cozart has his “mechanical adjustments” story, but if Barry Larkin turned a 29-year-old with a career 77 OPS+ into an offensive asset just by asking him if he’d ever considered crushing the inside of the ball, he deserves a second plaque in Cooperstown, this time as a coach.

The other two NL Central CSP fallers, Kolten Wong and Anthony Rizzo, are former top prospects who are now in their mid-20s and mashing. The fact that they’re seeing fewer hittable pitches should only add to the optimism. Just like last year, Ackley is slumping to start the season, but he rebounded in 2014, hitting .270/.308/.456 in the second half while delivering 10 of his 14 total homers — and pitchers seem to expect that clout to carry over. Kemp, who’s hitting .326/.357/.478 and hasn’t missed a game, qualifies again, which should make the Padres even happier to have him on the Dodgers’ dime. The presence of the two veteran Tigers is a mystery to me, but Eric Hosmer’s appearance just outside the top 10 makes sense: He hit well in the second half of last season, caught fire in October with everyone watching, and hasn’t slowed down in 2015. In a meaningful sample, pitchers have treated him more like the slugger he was always expected to be.

Now for the negative side. Here’s how far Ryan Howard has fallen: In 2009, the lefty’s called strike probability was 39.7 percent, third-lowest out of 221 hitters with at least 1,500 pitches seen. In 2015, Howard’s CSP stands at 50.8 percent, 22nd-highest among 233 hitters with at least 200 pitches seen. No one is worried about Howard’s power.

For the moment, Andrelton Simmons is hitting like an MVP candidate — which, for a guy with a glove as good as his, translates to “slightly above average.” If Simmons can continue to skirt the low BABIPs that leeched away his line in the last two years, avoid strikeouts, and contribute occasional power, he’ll be a superstar, and not only for his weekly Web Gems. Pitchers are pessimists on this point: Simmons has the highest CSP in baseball (57.3 percent). Another shortstop, Minnesota’s Danny Santana, was commonly cited as a regression candidate entering this season because his line last year was propped up by a .405 BABIP. His BABIP is still sky-high — it’s at .380 in 2015 — but even that’s not enough to make a hitter with a 22:0 strikeout-to-walk ratio imposing. Pitchers aren’t the slightest bit concerned, but Minnesota should be.

Curtis Granderson would be the list’s 11th man, following a sub-.400 slugging percentage last season and a .324 mark in 2015. Pitchers aren’t afraid of the 34-year-old’s dwindling power. Also of note: Pitchers have abandoned the Jason Kipnis bandwagon after his down 2014 and even less encouraging start to this season. Kipnis doesn’t quite make the top 10 risers in called strike probability — his was already high last year — but he does have the seventh-highest CSP, at 51.7 percent.

Which brings us back to Bryant. Since he wasn’t in the majors last season, we don’t have the data to compare him to his 2014 self, but we can compare his start to previous big-league debuts. The first 200 pitches to Bryant averaged a 40.8 percent called strike probability. Out of 617 players who’ve made their major league debuts and seen at least 200 pitches since 2008 (pitchers included), that’s the 32nd-lowest CSP over the first 200 pitches of a career — right next to Sandoval’s 40.7 percent. (Rizzo saw the lowest percentage, at 36.9.) Even compared to veterans, Bryant’s pitches seen tell a story of pitcher intimidation: Through Tuesday, his 41.5 CSP ranked 16th-lowest among all major leaguers with at least 200 pitches seen this season. I’ll spare you the Yogi quote, but my message is the same: If you want to know who the best hitters will be, just keep your eyes on the ball. The pitchers are trying to tell you.

Thanks to Rob Arthur, Jessie Barbour, Rob McQuown, Harry Pavlidis, and Nick Wheatley-Schaller for research assistance.