Price made a rapid ascent through the minor leagues in 2008, dominating High-A Vero Beach and Double-A Montgomery with sub-2.00 ERAs, 92 strikeouts in 91.2 innings, and a sparkling 11-0 win-loss record.  He was promoted to Triple-A Durham in August 2008 and held his own with a 4.50 ERA and 17 strikeouts in 18 innings.  He was called up to the big leagues and debuted with the Rays in a pennant race, pitching mostly out of the bullpen in September.  He pitched in three games in the American League Championship Series and two games in the World Series, striking out eight in 5.2 innings and acquitting himself well against October competition.

2009, however, was a year of struggles for Price.  He was sent back to the minors to begin the season, and after he was called up in late May, he posted a 5.60 ERA through his first 11 starts, failing to finish the fifth inning in five of those games.  Though he had posted 54 strikeouts in 53 innings, he had also given up 33 walks and 11 home runs.  He finished the remainder of the season on a stronger note, recording a 3.58 ERA over his final 12 starts, allowing only 6 home runs and 21 walks in 75.1 innings, but also cutting his strikeouts to 48 over that time frame.

In 2010, David Price emerged as the dominant starter people expected when he was drafted.  He finished third in the American League with a 2.72 ERA and boasted a splendid 19-6 win-loss record.  His peripherals were solid, with 188 strikeouts in 208.2 innings against 170 hits, 79 walks, and 15 home runs allowed.

Let’s take a look at the pitches that Price throws and how his repertoire has evolved during his tenure with the Rays.

Evolution of his repertoire

As you will see, classifying Price’s pitches can be a bit of a challenge when he’s frequently changing what pitches are in his repertoire.  However, I’ll go out on a limb and say that Price has used six pitch types in the major leagues: a four-seam fastball, slider, curveball, and three variations on a sinking pitch which I have identified as a two-seam fastball, split-finger fastball, and changeup.

I’ll explain the varieties of sinkers and changeups in a moment, as well as his journey from one pitch type to another.  Let’s start by talking about what Price threw when he broke into the majors in 2008, mostly pitching out of the bullpen.

For those of you not familiar with my previous writing, I enjoy analyzing the detailed pitch data that has been collected on pitchers for the past few years using the PITCHf/x camera systems that are installed in all 30 major league stadiums.  PITCHf/x tracks the speed and trajectory of nearly every pitch thrown in major league games, and the information is available for download from Major League Baseball’s Gameday website.  I have created a database of this pitch information and use it to analyze pitchers.  A number of other analysts have done similar work, including Josh Kalk, who published at the Hardball Times before being hired by the Rays in 2009.

Let’s look at what the PITCHf/x data show us about the pitch speed and spin movement of the pitches he has thrown.  We’ll go through each of the stages of the evolution of Price’s pitch repertoire, beginning with 2008.

David Price was basically a two-pitch pitcher as a reliever in 2008.  He threw a 94-mph four-seam fastball two thirds of the time, and the other third of the time he threw an 87-mph slider.  He also showed a changeup, but it was not a regular part of his repertoire out of the bullpen.

When he returned to the majors in May 2009 as a starting pitcher, he threw his four-seam fastball slightly slower, averaging 93.4 mph, but it was still his main pitch.  He expanded his off-speed offerings to right-handed hitters, adding an occasional changeup and a rare curveball or two-seam fastball.  You’ll recall that he was not especially effective during this period, and he struggled with his control.

In the second half of 2009, Price added a 90-91 mph two-seam fastball and cut back on his slider usage.  He threw the two-seam fastball almost exclusively to right-handed hitters, and used it almost as often to them as he did the four-seam fastball.

In his last three starts of 2009 and continuing into 2010, Price began to feature his curveball much more prominently and nearly abandoned his slider.  He also increased the speed differential between his four-seam fastball and two-seam fastball, as he threw the sinking fastball only 89-90 mph.

From May through August of 2010, the evolution of Price’s two-seam fastball continued.  He decreased the speed of the pitch even further to an average of 87 mph, such that it could probably best be described as a split-finger fastball.  At the same time, he upped his average four-seam fastball speed to 95 mph.

Finally, in August and September, Price largely abandoned the splitter (or whatever you want to call the high-80s sinking pitch) and replaced it with a harder two-seam fastball, thrown nearly as hard as his now 96-mph four-seam fastball.

We’ve examined each of the major phases of David Price’s pitch repertoire development. Let’s view the same thing a different way by looking at his pitch selection by game throughout his career.

We see that Price began his career as a four-seam fastball and slider pitcher, then added a two-seam fastball, then a curveball. Next, he replaced the two-seam fastball with a splitter, then went back to the two-seam fastball.  Toward the end of 2010, he increasingly favored the two-seam fastball over the four-seam fastball.

Pitch repertoire description

Let’s talk a little more about each of his pitch types.

First, the four-seam fastball.  Price has gained speed on his four-seamer in 2010.  In 2008, mostly in relief, his heat was measured at an average of 94.3 mph by PITCHf/x.  In 2009, it was measured at 93.5 mph; however, the Tampa Bay PITCHf/x system was recording speeds about 0.8 mph too slow that year.  That would imply Price’s real 2009 average fastball was more like 93.9 mph, assuming a roughly equal number of pitches at home and on the road.  In 2010, his average four-seamer was recorded at 95.2 mph.  The Tampa Bay PITCHf/x system was still recording speeds about 0.8 mph too slow until mid June; after that it was fairly well calibrated for speed.  The calibration adjustments to the PITCHf/x data give us an average four-seam fastball speed of about 95.4 mph for the 2010 season, or 1.5 mph faster than it was in 2009.

There has been some talk about Price progressively increasing his fastball speed during the 2010 season.  I believe this perception was mostly, but not totally, an artifact of the PITCHf/x calibration adjustment in mid June.  In August and September of 2010, his four-seam fastball averaged about 95.9 mph.

Price gets about eight inches of hop and seven inches of tail on his four-seam fastball.  That’s pretty typical for a major-league four-seamer.  He heavily relies on the pitch to left-handed batters, throwing it 75 percent of the time to them during 2010, whereas to righties he throws it only 49 percent of the time and mixes in the two-seam fastball.  The four-seamer is his primary strikeout pitch, generating 85 percent of his strikeouts of lefty batters in 2010 and 67 percent of his strikeouts of righties.  Price uses a classic four-seam grip, which you can see here.

We discussed earlier that David Price debuted a two-seam fastball in the second half of the 2009 season.  To recap, it was a pitch he threw 90-91 mph, about three mph slower than his four-seamer.  In 2010, he switched to a slower version that I’ve called a split-finger fastball before returning to a much harder two-seam fastball in August and September.  This new two-seam fastball averaged 95.1 mph, or within one mph of his four-seam fastball.  He gets about six inches of hop and 11 inches of tail on his two-seam fastball, which is about four or five inches of movement relative to his four-seamer.  In August and September, Price used the two-seamer only six percent of the time to left-handed batters but 37 percent of the time to right-handed batters.  You can see his two-seam fastball grip here and here.

Price’s split-finger fastball, as I’m calling it here, is a slower version of his two-seam fastball that existed primarily from May 12 to August 4, 2010.  His average speed on this pitch was about 87.4 mph, or almost eight mph slower than his four-seam fastball.  That’s almost a changeup speed differential, and I might have called it a changeup were it not for an even slower offspeed pitch type thrown by Price, which we will discuss in a moment.  His splitter gets about five inches of hop and nine inches of tail, fairly similar to his two-seamer.  During the time period where this was his main sinking offspeed pitch, May to July, he threw it 15 percent of the time to right-handed batters and rarely, if ever, to lefties. (I counted two such instances.) I’ve looked for game photos of his splitter grip from games in the May 12 to August 4 time frame, but so far I’ve had no luck finding any.  I’d guess his splitter grip is pretty similar to his two-seamer grip, with the variation coming in how tightly he grips the ball into his palm rather than from major changes in seam orientation.

As mentioned previously, Price has another slower sinking off-speed pitch.  I’ve called this pitch type a changeup.  His changeup averages about 83 mph, or 12 mph slower than his four-seam fastball.  He gets about seven inches of hop and 10 inches of tail on his changeup, very similar spin movement to his two-seam fastball.  He has thrown the pitch exclusively to right-handed opponents, about four percent of the time during the 2010 season.  I’ve not been able to find a game photo that I can conclusively identify as a changeup grip, but this one looks suspiciously like a circle change to me, though it wasn’t taken from the best angle for grip identification.

Coming up through the minors, Price’s slider drew high praise, and it was his main secondary pitch at the end of the 2008 season.  However, we’ve seen a lot less of the slider since he developed the curveball toward the end of 2009.  He throws the slider about 86-87 mph, with very little spin movement.  When it was his main breaking ball in 2008-2009, he threw it equally to righthanders and lefthanders.  In 2010, he used it mostly to lefties (12 percent of the time) and rarely to righties (three percent of the time), preferring instead to use the curveball against right-handed batters.  You can see a photo of his slider grip here.

There is some indication in his last couple starts, September 23 and September 28, 2010, that Price might be evolving his now seldom-used slider into a cutter.  The slider was thrown a little harder with a little more hop in those two games.  It’s a very small sample size, a total of five pitches, but it was enough to pique my curiosity, and it’s something I’ll be paying attention to in the playoffs.

Price’s favorite breaking pitch is now a curveball. He throws the curveball at about 78 mph on average, and he gets five inches of cut and eight inches of drop due to spin.  If you include the effect of gravity, his curveball drops about 2.5-3 feet. In 2010, he used the curve 11 percent of the time to left-handed batters and 17 percent of the time to right-handed batters.  He uses a spike curveball grip that you can see here.

Pitch mix

We’ve looked at some changes in Price’s pitch repertoire and examined some of how he uses each of his pitches.  Let’s also look at how he mixes his pitch types in different ball-strike counts.  Below, I show his pitch mix for 2010, split out by batter handedness.

To left-handed batters, Price uses mostly his four-seam fastball.  This is especially true when he falls behind in the count; lefties will see the four-seamer 87% of the time in this case.  When he is even or ahead in the count, he’s a little more likely to throw a breaking pitch (26% of the time).

Price uses his off-speed and sinking pitches more freely with right-handed opponents, though he still favors his four-seam fastball as a strikeout pitch on 0-2 (70% of the time).   He throws his curveball most often (19% of the time) to righties when even or ahead in the count, and his splitter and two-seam fastball when he gets behind in the count.

Pitch results

What kind of results did Price get with each of his pitch types in 2010?

Price has thrown all six of his pitch types for strikes this year.  The four-seam fastball and two-seam fastball generated impressive whiff rates, 11% and 9% respectively, as compared to 7% and 6% major-league average.  His off-speed and breaking stuff, however, was not terribly impressive in that regard.  He had some pretty low batting averages allowed on balls in play (BABIP) for the two-seam fastball, the changeup, and the slider.  I looked a little deeper into this but did not find any patterns that suggested this was likely to be a persistent trend.  It’s possible I missed something that data from additional seasons may reveal more clearly.

Let’s look now at where he locates his pitches to right-handed and left-handed batters. Strike-zone location charts are shown from the perspective of the catcher. The location of a pitch is indicated where it crossed the front plane of home plate.

The first thing that is obvious from this graph is that David Price faces many more right-handed batters than left-handed batters.  Price has faced 77 percent right-handed batters; this is fairly typical for a left-handed starting pitcher.  Against righthanders, Price throws his four-seamer both inside and outside.  He’s very successful at getting swinging strikes when he gets the ball up and away within the strike zone.  Against lefthanders, he keeps the four-seamer away, and he gets quite a few called strikes where the umpires typically expand the outside edge of the zone to lefty batters.

Price pitched with a similar pattern with all of his sinking pitches: the two-seam fastball, the split-finger fastball, and the changeup (as I have labeled them), so I’ll discuss them together.  He rarely throws any of the them to left-handed opponents.  To right-handed batters, he kept them on the outside edge.  He does well at throwing them for strikes, and the two-seamer seems to be the best of the three at fooling opponents into swinging and missing.

As already discussed, Price cut back on his slider usage in 2010.  When he did use it, he kept it away from lefties and brought it inside to righties, which is a typical approach for a left-handed pitcher.  Price was able to throw the slider for strikes, but otherwise it was fairly unremarkable, generating below-average numbers of whiffs.

On the other hand, Price used the curveball much more like a typical breaking pitch, inducing swings and misses on pitches down and out of the zone, though again, his whiff rate on the curve was below average when compared to the league.  To right-handed batters, he mostly kept the curveball across the middle of the plate, and he was able to keep it in zone at a slightly above average rate (62 percent vs. league average of 57 percent).  Against left-handed batters, he moved the curveball more toward the outside edge, and consequently, his curveball strike rate was lower to lefties (57 percent).

Conclusions

It has been interesting to watch David Price develop into the ace of the Rays’ pitching staff.  It has been fascinating for me to dive into how he accomplished that and to learn about the evolution of his repertoire across seasons.  It’s rare that a pitcher is that willing to continuously experiment, adapt, and evolve.

He has a dominant 95-96 mph four-seam fastball that serves him well against left-handed batters and a four-seamer/two-seamer pairing that has helped him improve against right-handers in 2010.  His breaking pitches are average offerings that fill out a solid repertoire.  Though his performance on batted balls going forward may not be expected to be as good as it was in 2010, his stuff certainly seems to support his ability to carry the mantle of staff ace in the future.

 In Part 1 of this series, we examined Brian Bannister’s suggestions for why he has been able to beat the league BABIP. He indicated that it was probably due to pitching more often in favorable pitcher’s counts and inducing balls in play with two strikes, when the hitter is against the ropes. However, the evidence didn’t show much advantage for Bannister. We noted that he did pitch a little more often in favorable counts, but this led to him avoiding walks more than anything; it had little salutary effect on his BABIP.

In Part 2 of this series, we learned about the pitches that Bannister threw during 2007 and how he used them. We saw that the fastball and curveball were good pitches against right-handed hitters, and the slider was a good pitch against left-handed hitters.

Part 1
Part 2
Part 3

In this final part of the series, we’re going to marry those two approaches to see if we can uncover any patterns that might explain Bannister’s BABIP performance. In this portion, I’m not concentrating so much on evaluating Bannister’s own statements, as I did on Part 1. Rather, I’m thinking more about what we can expect from Bannister in the future. I’m also interested in investigating techniques that could prove useful for evaluating DIPS theory on a component basis as we accumulate more PITCHf/x data in the coming seasons.

Should we expect Bannister to maintain any of his BABIP edge and thus his 3.87 ERA from 2007? Or are the projection systems like PECOTA (subscribers only) and CHONE more reasonable when they project an ERA of 5.19 or 4.74?

I recently received my Hardball Times 2008 Season Preview book in the mail, and a few quotes from Bradford Doolittle’s article on the Royals encapsulate the factors that are driving the projections for Bannister:

He got lucky in terms of keeping fly balls in the park, has a low strikeout rate and has a poor groundball rate. Look out.

…when looking at measures like average on balls in play, groundball percentage and homers/fly ball, it’s pretty apparent that he was a beneficiary of good fortune last season. His list of comps isn’t encouraging, either. He’s the most likely player on the Royals to suffer a collapse in 2008.

Mainly, I will focus on the question of Bannister’s batting average on balls in play (BABIP), but let me briefly address the other concerns that Doolittle raised.

Bannister got lucky in terms of keeping fly balls in the park? Yes. Or at least his HR/flyball was lower than average, to the tune of about 5 less home runs allowed than “expected” in 2007. I’m still agnostic as to whether this is due primarily to luck or skill for every pitcher. This probably bears some investigation in Bannister’s case, but it’s a less significant factor to his run prevention than his BABIP performance, so I’m not going to examine it further in this series.

What about his ground ball percentage? As far as I can tell, it’s fairly close to normal, 42% of balls in play, compared to league average of 43%. THT and I must be using slightly different definitions, however, since they put Bannister’s number at 41% and the league at 44%. (My data comes from MLB Gameday; THT gets theirs from Baseball Info Solutions.) Ground balls are much less likely to turn into extra base hits than are fly balls or line drives. I’ll talk a bit more about this as we delve into the BABIP numbers, but given my numbers for balls in play, I wouldn’t tab Bannister’s ground ball percentage as a particular problem area.

Bannister does have a low strikeout rate, down in Chien-Ming Wang and Jon Garland territory in 2007. With an 89-mph fastball and no killer secondary pitch, that’s not likely to improve. It’s a real area for concern and the reason there is so much focus on his BABIP. If you don’t strike ‘em out, you care a lot more about what happens when they put it in play.

That brings us back full circle to BABIP. If it wasn’t getting balls in play on favorable counts that caused Bannister’s outstanding BABIP performance in 2007, what was it? We’ve already taken a look in Part 2 at Bannister’s performance by measuring the average run value of each of his pitch types. I also listed the BABIP against each pitch, split out against left-handed hitters and right-handed hitters. From the tables in Part 2, you can infer which pitch helped his BABIP the most, but let’s show that explicitly as a starting point here. I also want to note again that the PITCHf/x data set covers only about half of Bannister’s pitches during 2007, so these BABIP numbers broken down by pitch type will not add up to his full season BABIP numbers.

You can see that Bannister’s BABIP advantage primarily came against right-handed hitters; against lefties he’s pretty close to the league-average mark. We have PITCHf/x data for 263 of his 538 balls in play, and his BABIP of .255 for that sample matches up pretty well with his full season BABIP of .262.

I’m a little concerned here about drawing conclusions from too small a sample size. I’ll focus on the 12-for-71 performance of right-handed hitters against Bannister’s fastball, the cause of 10 of the 12 “missing” hits we’d have expected Bannister to allow in our sample, out of 23 total hits missing from Bannister’s full 2007 season, if he’d allowed a league-average BABIP.

If I understand my binomial distribution statistics correctly, and it’s been about ten years since I last dusted them off, we can state at the 99% confidence level that Bannister’s BABIP against righties with the fastball was not merely the product of chance. Only one of the other BABIP splits meets even the 90% confidence level, that being the 2-for-18 by righties against the curveball. Even though it’s a smaller sample with less confidence, we’ll take a brief look into it, too.

So, the $64,000 question–or in Bannister’s case, perhaps it’s a $364,000 question–how did Bannister manage to get righties to hit a paltry .169 on balls in play against his pedestrian-looking 89-mph fastball? Let’s start by breaking that down by type of ball in play:

3-for-32 on fly balls
0-for-17 on ground balls
9-for-11 on line drives
0-for-11 on popups

There are at least two ways to examine this data. We know that different types of balls in play have different expected BABIP. Mitchel Lichtman published the seminal work on this topic, looking at data from 1992-2003. My numbers from 2007, and my batted ball categories, differ a little from MGL’s work, but we’re in fairly good agreement.

First, what is the BABIP for different batted ball types? Here are my numbers for the major leagues from 2007:

0.169 on fly balls
0.251 on ground balls
0.729 on line drives
0.025 on popups

(As an aside, why do people prefer ground ball pitchers if the BABIP on ground balls is higher than on that on fly balls? It’s because 78% of fly ball hits go for extra bases, whereas only 9% of ground ball hits go for extra bases.)

Second, what is the expected distribution of balls in play to the different types, more to the point, what is the expected distribution against the fastball? Here I’ll draw on one of John Walsh’s many excellent pieces on PITCHf/x. According to his numbers, the typical fastball produces 31.5% fly balls, 38.8% ground balls, 19.9% line drives, and 8.7% popups.

Let’s start with the second point. Maybe Bannister excelled with his fastball to right-handed hitters because he induced a very favorable mix of balls in play? His breakdown in our sample was 45% fly balls, 24% groundballs, 15% line drives, and 15% popups. A quick look shows us that he’s getting more fly balls and popups and less ground balls and line drives. Based on his batted ball mix alone and assuming a league-average BABIP, we would have expected Bannister to allow 18 hits in 71 at bats against the fastball. In other words, it explains almost 40% of the difference. Definitely interesting.

Again, I’ll caution that this is a small sample size. MGL’s work showed that the mix of balls in play is a much more repeatable skill on the part of a typical pitcher than is their overall BABIP, particularly with regard to ground balls and fly balls, but also to a fair extent with line drives and infield popups. However, I’m not sure we really know how to regress these numbers to the mean. I suspect the batted ball mix we see from Bannister is not mostly due to chance, but we don’t really have the studies to prove it yet one way or the other at the pitch-type level.

But Bannister only allowed 12 hits against his fastball to righties, not 18, so what about the remaining 6-hit difference? Given his batted ball mix, what would we have expected hitters to hit?

5.4-for-32 on fly balls
4.3-for-17 on ground balls
8.0-for-11 on line drives
0.3-for-11 on popups

The actual numbers on line drives and popups are close enough to the expected values that I’m going to ignore them. But Bannister saved 4+ hits on ground balls and 2+ hits on fly balls. Why? It’s instructive to look at where the 17 ground balls went. Three went to third base, 12 to the shortstop, and 2 to second base. Who played shortstop for the Royals last year? None other than defensive wiz Tony Pena, Jr. I don’t know whether Bannister did something special to get all the right-handers to hit his fastball on the ground toward shortstop, but it certainly turned out well for him. We can probably expect some regression in this area next year, but there’s also a hint that Bannister is maximizing the strengths of the defense behind him.

Looking at the fly balls, there’s also an obvious pattern to where they were hit. When right-handed hitters put the fastball into play in the air, it goes toward right field. Take a look at where the 32 fly balls landed on this ball-in-play chart (which also charts the two home runs hit by righties off the Bannister fastball in our PITCHf/x sample). Field dimensions are approximate.

The center fielder David DeJesus is fielding balls into right center field, and the right fielders (mainly Mark Teahen and Emil Brown) are mainly fielding balls from straightaway right field toward the right field line.

Here’s the chart for all types of balls in play by righties against the Bannister fastball.

The line drives follow a similar pattern to the fly balls, but since they’re hit harder, most of them fall in for hits. You can see the ground ball cluster around the shortstop. The popups Bannister induces are mostly foul balls on the first base side.

I’m not quite sure what to make of these patterns of balls in play. Bannister probably won’t get quite as lucky in the placement of balls relative to his fielders, but I don’t see any particular warning signs that scream, “Look out”, as the Hardball Times season preview put it. However, without having looked at more pitchers than just Bannister, I wouldn’t put much stock in my ability to notice any such indications.

There are other factors that could come into play. Did Bannister face weaker than average hitters? The overall answer to that question is no. Bannister’s average opponent sported a .267/.339/.424 line (avg/obp/slg), compared to the average opponent faced by the average AL pitcher at .268/.336/.423. There’s not much difference, certainly nothing that would help explain Bannister’s BABIP. What about the average right-handed hitter to whom Bannister threw a fastball? We know that weaker hitters tend to see more fastballs. Was this also true in Bannister’s case?

To answer this question, I’ll include the two home runs (hit by A-Rod) along with the 71 balls in play. In our sample, Bannister faced 37 different hitters, and I’ve weighted their performance according to the number of times each hitter faced him, for a composite fastball opponent hitting .273/.338/.436. That’s a slightly better hitter than average–there’s no sign that Bannister improved his fastball BABIP by feasting on weaker hitters. Even if you take A-Rod’s two at bats out of the equation, the numbers are still a hair above average.

I also looked at location of fastballs in the strike zone relative to what happened to the balls in play by right-handed hitters. The only thing I noted was that the low outside corner was a good place for Bannister to generate ground balls. The heart of the plate was the location for the two fastballs that A-Rod launched for home runs, but up and over the middle was also the place where Bannister generated a lot of foul pop flies.

What about the effect of previous pitches? Righties went 0-for-9 on first-pitch fastballs put into play. They went 6-for-34 (.176) following a previous fastball, 1-for-8 (.125) following a slider, 3-for-10 following a changeup, and 1-for-10 following a curveball. I could look at pitches other than the one immediately prior to the pitch put in play, but there doesn’t seem to be much of a pattern here.

We already determined in Part 1 that Bannister gained only a very small advantage overall from pitching in favorable counts. In Part 2 we learned that Bannister tended to use the fastball with 0 or 1 strike. From those two pieces of information, there’s nothing that suggests that Bannister got good results on balls in play with the fastball against right-handed hitters because he was throwing it to them in favorable counts, but we ought to take a look at that specific combination just to make sure. On what counts did righties put the fastball into play?

It turns out there’s not much to see from that angle, either. In the four counts most favorable to the pitcher (0-2, 0-1, 1-2, 2-2), the righty hitters were 6-for-28 (.214) against the fastball, and in the four counts most favorable to the hitter (3-1, 2-0, 3-2, 1-0), they went 4-for-25 (.160). Those pitcher’s counts typically result in more balls in play than the hitter’s counts; conversely, pitchers throw more fastballs in hitter’s counts, although not quite enough to balance out the first effect. So Bannister was near the average in terms of his mix of counts, and he actually got slightly better results on the fastball put in play on a hitter’s count than on a pitcher’s count.

I also promised a look at one other pitch type–why did righties hit 2-for-18 off of Bannister’s curveball? We have a much smaller sample, but let’s look at a few of the same metrics that we did for the fastball. How did righties hit on different types of batted balls against the curveball?

0-for-7 on fly balls
1-for-9 on ground balls
1-for-1 on line drives
0-for-1 on popups

The expected distribution of balls in play to the different types against the curveball is 25.2% fly balls, 48.1% ground balls, 18.5% line drives, and 6.8% popups. Did Bannister induce a favorable mix of balls in play with his curveball to right-handed hitters? His breakdown in our sample was 39% fly balls, 50% groundballs, 6% line drives, and 6% popups. He’s moved a couple line drives to the fly ball column, which is good and accounts for at least one of the three “missing” hits, but we’re so far into the realm of small sample size that we could be looking at things like scoring decisions about how to classify the batted balls as much as anything about Bannister’s pitching.

I’ll show you the ball-in-play chart for right-handed hitters against his curveball, but I’m not going any further down this path since the sample size is so small.

In summary, Bannister’s BABIP performance appears to be partly luck and partly skill. He’ll probably have a few more fly balls fall in for hits than he did in 2007, but he also induces a favorable mix of batted balls and seems to use the strengths of the defense behind him well. It’s not clear how well that will carry over.

I’d be skeptical of alarmist claims that Bannister is simply going revert completely to the league mean for BABIP. He seems to know how to use his 89-mph fastball to its best advantage, and the existing batted ball research doesn’t really know how to deal with this kind of information yet. However, repeating his league-leading .262 BABIP is also an unreasonable expectation.

As a Royals fan, I’d love to see him develop stronger secondary pitches and to continue to hone his craft in a Maddux-like fashion. It’s exciting when a major-league pitcher even knows what BABIP is. One of my favorite things about the game of baseball, perhaps my very favorite, is the game of wits that goes on between pitcher and batter. Bannister seems well-equipped for that game. I’m cheering for him in 2008 and wish him the best, but I’m not going to go out on a limb and predict his ERA. I don’t know enough to tell you that kind of thing.

I do hope that with another year of PITCHf/x data, we can gain a better understanding of the interaction between particular pitches and results on balls in play, and a result, a better ability to quantify the skills that make pitchers successful.

Part 1
Part 2
Part 3

In Part 1 of this analysis, we examined the league numbers for batting average on balls in play (BABIP) and whether Bannister was able to beat the league BABIP by pitching in favorable counts. We found that he did not gain any particular advantage by inducing more balls in play on two-strike counts, so we turn elsewhere to seek an explanation for his 2007 performance.

Part 1
Part 2
Part 3

What pitches does Brian Bannister throw? The scouting reports tell an interesting tale, especially if you follow them back a couple years. In the minor leagues, the cut fastball was reputed to be his best pitch. His four-seam fastball was thrown in the high 80′s, touching 90, although he was able to locate it well, his curveball was a big breaker that was considered a plus pitch, his changeup was a work in progress, and his slider was regarded as a pitch likely to be scrapped. But in the fall of 2006 in the Mexican League, Bannister worked on a two-seam fastball, and after joining the Royals in trade for Ambiorix Burgos, he scrapped his cutter, experimented with different speeds on his curveball, and started throwing a slider again.

What can we see in the PITCHf/x data regarding his pitch repertoire in 2007?

As far as I can tell, Bannister basically threw four pitches in 2007. I couldn’t find much evidence of a cutter, so I believe he didn’t throw it very often. I couldn’t tell his four-seam and two-seam fastballs apart, so I’ve lumped them together as fastballs. Often, I can see evidence of two clusters when pitchers throw both types of fastballs, even if the exact delineation is hard to make. In Bannister’s case, I didn’t even see two separate clusters. In addition to the fastball, he threw a changeup, slider, and two varieties of curveball. Let’s examine these pitches in more detail.

Bannister’s fastball runs 87-91 mph, and the average spin deflection he gets on the fastball is a 9-inch hop and a 1-inch tail in toward right-handers. Compared to a league-average fastball, that’s 2 mph slower and with about 5 inches less lateral movement. The fastball is Bannister’s main pitch to both lefties (52% of the time) and righties (56%).

His changeup runs 82-86 mph; the five-mph separation between his fastball and changeup is quite a bit less than the league average separation of nine mph between those pitches. The average spin deflection on his changeup is a 2-inch hop and a 6-inch tail in toward right-handers. Bannister uses the changeup fairly often to lefties (19% of the time), but it’s his fourth pitch to righties (only 9%).

Bannister’s slider runs 84-87 mph, and the average spin deflection he gets on the slider is a 3-inch hop and a 3-inch break away from right-handers. That’s a couple inches more lateral deflection than a league-average slider. He uses the slider almost equally to lefties (15% of the time) and righties (20%). It’s somewhat unusual to see a right-handed pitcher throw the slider that often to lefties. Reference this quote from Bannister after his June 29 start against the White Sox:

Bannister rarely faced any trouble the rest of the way. He entered the night with a .299 average versus left-handers, 83 points higher than righties. The right-hander’s goal was to equalize those numbers, and he incorporated a new slider to great effect.

“It’s a new pitch,” Bannister said. “I used to throw a lot of cutters. I have been working on it all year, and tonight, I threw it a lot more. … Mac has really worked with me on things and I think the slider really helped, especially against the left-handers.”

Bannister’s fourth pitch is a curveball, and it also presents something I haven’t seen much with other pitchers. For most of the year, Bannister threw a slow curveball, running 70-78 mph, and there’s some indication he was even purposefully varying speeds within that range. Then, in his August 17 start against Oakland, he changed curveballs, unveiling a harder curve running 77-82 mph, with about five inches less drop than his curveball from earlier in the year. It seems to have been a complete switch from one to the other, since I don’t see any game in which he threw both varieties of curveball. Overall, Bannister threw the curveball 14% of the time to both lefties and righties.

Here’s another graph showing the movement due to the forces of spin deflection and gravity on his pitches in the last quarter-second before they cross the plate, and here are a couple other ways to look at the vertical vs. horizontal deflection over the whole pitch trajectory:

Next, let’s look at how Bannister mixes his pitches in different ball-strike counts, which I’ve split out by batter handedness. Bannister is more consistent than most pitchers about throwing all of his pitches across all counts, but he still shows some tendencies by count.

To righties, he likes to start out with a fastball or an occasional slider to get ahead of the hitter; both are pitches he can throw for strikes. The changeup features most often at 1-strike counts (17%), compared to only 6% of pitches with 0 or 2 strikes. Like most pitchers, he’ll throw the curveball when ahead (24%) but rarely when he’s behind in the count (3%).

To lefties, he also likes to start out with the fastball, and he’ll throw the curveball at 0-0 (17%) or when ahead in the count (26%) but not when behind (4%). By contrast, the slider rarely gets thrown at 0-0 or 1-0 (5%), but he likes to throw it with 1 or 2 strikes (23%).

He doesn’t seem to have a particular favorite strikeout pitch against either righties or lefties. The fact that he doesn’t have one great pitch to rely on for strikeouts may go a long way toward explaining why he doesn’t get that many of them. Along those lines, let’s examine the results that Bannister gets with each of his pitches. I’ve split the curveball into the slow curveball (Curveball1) and the late-season hard curveball (Curveball2) since he got quite different results from the two.

A good way to give an overall grade to Bannister’s pitches is to use the linear weights run values like we did earlier. In this way it’s easy to see that the fastball is Bannister’s best pitch; with it he can shut down righties and hold his own against lefties. The slider is a strong pitch against lefties but weak against righties, and the curveball is the opposite.

The league average information comes from John Walsh’s article. Now, let’s look at how Bannister locates his pitches and in more detail at his results.

Against lefties, Bannister pounds the outside part of the plate, getting a lot of called strikes and some foul balls on the outside edge. When he misses down out of the zone or too far outside off the plate, lefties rarely swing and miss. When he gets his fastball over the inside half of the plate, lefties make good contact. They hit .338/.615 (avg/slg) on balls in play, compared to league average of .330/.521. In the 65 fastballs to lefties tracked by PITCHf/x that were put in play, 4 of them went for home runs.

Against righties, Bannister lives in the zone and gets surprisingly good results for an 89-mph fastball. His called and swinging strikes and foul ball rates are all about normal. However, when righties put the fastball in play, they hit only .192/.356. We’ll examine this in more detail later since it seems to be a large component of Bannister’s success.

Bannister throws the slider for strikes to both lefties and righties, throwing 68% strikes compared to league average of 64% for the slider. He pitches it down and in to lefties, and when he gets it right on the corner, he has a good shot at get a swing and a miss. Otherwise, it’s likely to be fouled off, 27% of the time by lefties, compared to 17% league average. When lefties do put the slider in play, mostly groundballs, they hit a weak .200/.320, compared to .310/.481 off a league average slider.

Righties fare better against the slider. He doesn’t get many swings and misses nor as many foul balls. In fact, righties put the slider in play 30% of the time. Bannister works down and away a lot, but he doesn’t look terribly effective there. He either misses the strike zone for a ball or gets a little bit of the plate and lets the batter make contact. Righties hit .368/.632 against the slider when they put it in play, including 3 home runs out of 38 balls in play in the PITCHf/x data set.

Bannister has a harder time throwing the changeup for strikes than his other pitches, getting a strike 60% of the time, which is in line with league average for the change. To lefties, he works away and down. He doesn’t get many swings and misses. He gets some called strikes on the outside part of the plate. When lefties get the ball in play, they hit .235/.471, compared to league average against the changeup of .319/.502. The change is only an average pitch for him to lefties because he can’t generate more whiffs with it.

To righties, Bannister works the changeup inside, generating a lot of foul balls, and down, mostly below the knees for a ball. Those two events cover 78% of his changeups. He doesn’t get many strikes, either called or swinging, only a mere 5%, compared to a league average of 24%. That’s what makes this pitch somewhat below average for him because when hitters put it in play, they hit .300/.400, which is slightly better than average.

During the first two thirds of the year, Bannister used a slower curveball, and he almost never threw it for a strike to lefties, staying well down or well away from the zone, for 71% called balls. He also threw it down off the outside corner for a ball to righties fairly often, but he was more willing to get in the zone for a strike to them, and he had fairly good results when he did, including holding opponents hitless on 7 balls in play. The slow curveball grades out as a plus pitch to righties and minus pitch to lefties, but it’s in a small sample and mostly has to do with hitting the strike zone or not.

During the final third of the year, Bannister switched to a harder curveball and was more willing to challenge the hitters with it, particularly lefties, who went from seeing 29% strikes with the slow curve to 64% strikes with the harder curve. He still worked down but now was willing to come across the outer half of the plate with the curve. It garnered him a few swings and misses, whereas no lefty whiffed on the slow curve. However, it didn’t end up being any improvement for Bannister since lefties pounded the new curveball for 7 hits in 15 balls in play. Again, it’s hard to say with the small sample size whether the ball in play results for the curve have any significance.

To righties, the pitching pattern with the hard curve was similar to that with the slow curve. Again, they didn’t have a whole lot of luck on balls in play, netting 2 singles in 11 attempts, mostly off pitches on the outside edge. Once again, small sample size prevents us from drawing strong conclusions.

Combining the ball in play performance for the two types of curveballs yields slightly bigger samples–righties hit 2 singles in 18 tries, lefties knock 6 singles and 3 doubles in 21 attempts. I’m still not sure if the curve is really more effective against righties; in any case, Bannister doesn’t throw it often enough to make a huge impact on the overall ball-in-play numbers we are investigating.

Next, let’s look at Bannister’s pitch sequencing. Here’s a table showing what pitch a hitter is most likely to see from him based on what the previous pitch was.

To righties, Bannister doesn’t have many discernible patterns in how he mixes his pitches, although he doesn’t throw the changeup to them much following a breaking ball. To lefties, there are a couple more noticeable patterns. He likes to follow a changeup with a fastball, and he doesn’t throw his curve much after a slider or changeup.

Now that we know a bit more about Bannister’s repertoire and how he employs it, we’re ready to move on to Part 3 of the series, which will examine the reasons Bannister’s 2007 performance on BABIP was better than the league and whether we can realistically expect those things to be repeated in the future.

Part 1
Part 2
Part 3

I’ll warn you from the start that the title is a tad ambitious. I don’t know exactly how Brian Bannister wins in the major leagues with a below-average fastball speed, but I hope to share some of what I have learned on the topic. This article will take the form of a three-part series.

Part 1
Part 2
Part 3

In case you’ve been hiding under the proverbial sabermetric rock the last few weeks–maybe you’re one of those weirdos who believe players are human or you’ve been out of your garage recently to look at the sky–Brian Bannister gave a fascinating three-part interview to Tim Dierkes at MLB Trade Rumors last month.

In Part 3 of the interview, Bannister talked about his opponents’ batting average on balls in play (BABIP).

I think a lot of fans underestimate how much time I spend working with statistics to improve my performance on the field. For those that don’t know, the typical BABIP for starting pitchers in Major League Baseball is around .300 give or take a few points. The common (and valid) argument is that over the course of a pitcher’s career, he can not control his BABIP from year-to-year (because it is random), but over a period of time it will settle into the median range of roughly .300 (the peak of the bell curve). Therefore, pitchers that have a BABIP of under .300 are due to regress in subsequent years and pitchers with a BABIP above .300 should see some improvement (assuming they are a Major League Average pitcher).

Because I don’t have enough of a sample size yet (service time), I don’t claim to be able to beat the .300 average year in and year out at the Major League level. However, I also don’t feel that every pitcher is hopelessly bound to that .300 number for his career if he takes some steps to improve his odds – which is what pitching is all about.

In the interview, Bannister postulated a reason for his success on BABIP.

So, to finally answer the question about BABIP, if we look at the numbers above, how can a Major League pitcher try and beat the .300 BABIP average? By pitching in 0-2, 1-2, & 2-2 counts more often than the historical averages of pitchers in the Major Leagues. Until a pitcher reaches two strikes, he has no historical statistical advantage over the hitter. In fact, my batting averages against in 0-1, 1-0, & 1-1 counts are .297/.295/.311 respectively, very close to the roughly .300 average.

My explanation for why I have beat the average so far is that in my career I have been able to get a Major League hitter to put the ball in play in a 1-2 or 0-2 count 155 times, and in a 2-0 or 2-1 count 78 times. That’s twice as often in my favor, & I’ll take those odds.

This interview has gotten a lot of buzz in sabermetric cyberspace. Several people have taken a look at BABIP at different ball-strike counts, including my colleague at StatSpeak, Pizza Cutter. There seems to be some ability for the pitcher to control the count on which hitters put balls into play, but it looks like a fairly small effect on average. (Pizza, correct me if I’m summarizing your conclusions incorrectly.)

Bannister also mentioned to Dierkes that getting two strikes on the hitter gives him the strategic advantage in terms of pitch selection.

It is obvious that hitters, even at the Major League level, do not perform as well when the count is in the pitcher’s favor, and vice-versa. This is because with two strikes, a hitter HAS to swing at a pitch in the strike zone or he is out, and he must also make a split-second decision on whether a borderline pitch is a strike or not, reducing his ability to put a good swing on the ball. What this does is take away a hitter’s choice. If I throw a curveball with two strikes, the hitter has to swing if the pitch is in the strike zone, whether he is good at hitting a curveball or not. He also does not have a choice on location. We are all familiar with Ted Williams‘ famous strike zone averages at the Baseball Hall of Fame. It is well-known that a pitch knee-high on the outside corner will not have the same batting average or OBP/SLG/OPS as one waist-high right down the middle. Here is a comparison of the batting averages and slugging percentage on my fastball vs. my curveball:

Fastball: .246/.404
Curveball: .184/.265

We do know from John Walsh’s work something about batting average and slugging percentage against the typical major-league fastball (.330/.521) and curveball (.310/.471). If Bannister is correct in his numbers, he’s doing quite a bit better than the league with both the fastball and curveball. But is Bannister correct in the numbers he quotes and assertions he makes?

So far, most people are accepting what Bannister said at face value. Let’s take a closer look and see if we should believe his numbers and conclusions. We’ll draw on two data sets from the 2007 season. One is the standard pitch-by-pitch result data for all of Bannister’s 2603 pitches in 2007. With this data set we can examine results on balls in play and how Bannister performed in various ball-strike counts. The second data set is the detailed PITCHf/x trajectory data recorded for 1304 of Bannister’s pitches, or about half of his starts. With this data set we can identify pitch types and reliable strike zone location information in order to gain a greater understanding of Bannister’s pitching strategies.

First, what does it mean when we say that Bannister had abnormal success with BABIP in 2007? The work of Voros McCracken and others over the last decade has shown that pitchers in general have limited control over the results of balls in play, such that on a single-season level, chance plays a bigger role in batted ball outcomes than does the pitcher’s own skill. To oversimplify the argument, a pitcher who performed better than the league average for BABIP is considered to have gotten lucky and should be expected, on average, to have a BABIP next year that is much closer to the league average BABIP than to his own performance in BABIP from the preceding season.

The American League BABIP in 2007 was .304. (I’m including sacrifice flies as outs in my BABIP calculations.) Brian Bannister had the lowest BABIP among starting pitchers in the American League at .262. If Bannister had performed at the league average for BABIP, he would have allowed 164 hits on balls in play instead of the 141 he actually allowed. Allowing 23 extra hits on balls in play would translate into roughly 15 extra runs, or 0.80 runs per 9 innings. In other words, if you attribute the superior BABIP performance entirely to luck, when predicting how Bannister will do in the future, you’d be better off acting as if he had a 2007 ERA more like 4.60 rather than the 3.87 mark he actually posted.

Researchers familiar with this topic will know that we can’t entirely attribute a pitcher’s BABIP performance to luck. Any given major league pitcher has some control over BABIP. However, since that control is small, we regress BABIP performance heavily to the mean when using BABIP data for making predictions of future performance. We might also want to take into account the quality of the fielders that played behind the pitcher or the park in which they played. I’m not going to go into detail on either of those two fronts since those topics have been covered thoroughly by others more able than me, but suffice it to say that neither effect comes close to explaining Bannister’s BABIP performance in 2007. From Bannister’s 2007 sample, regressed to the mean, we might estimate his BABIP skill at a performance level .006 better than league average. The Royals fielders (1379/4493) in 2007 were just slightly below average at converting balls in play into outs, with a team BABIP of .307. So we’re still left wondering how Bannister managed a BABIP of .262.

The B-A-B-I-P
Yes, that’s the stat for me!
It stands alone as the stuff of luck
The B-A-B-I-P.

What of Bannister’s proposition that it’s all in the ball-strike counts in which he pitches and induces balls in play? Does he actually pitch in favorable counts and get balls in play in those counts more often than other pitchers? If he does, does that explain his BABIP performance, or are we still left attributing most of it to chance?

Before we can answer those questions, we need to establish a method for measuring a pitcher’s (or the whole league’s) performance at a given count. The most accurate and comprehensive way is to determine the run values of various events using linear weights. For sake of brevity, I won’t go into the details of determining the run values for different counts in this article. You can find a lot of good material on the subject at Tango’s Book blog, this thread on Joe Sheehan’s recent article being one example.

Here are the 2007 American League average run values for a pitch thrown at each count. The values for a ball, a strike, or a ball in play at a given count are shown relative to the overall run value for that count. (Strikes here include foul balls at a two-strike count.)

You can see that the counts overall most favorable to pitchers are 0-2 and 1-2, followed by 0-1, 2-2, and 1-1. When a ball is put in play, every count favors the hitter on a net basis except 0-2. When adjusted for expectations at that count, the hitter also comes out behind at 2-0, 3-0, and 3-1. Why does putting the ball in play at these counts favor the pitcher? Because the hitter is fairly likely to get a walk or bat in another hitter’s count if he takes the 2-0 pitch. The hitter does better than the average for all counts when putting a ball in play at 2-0, but not better than the average of the other results he gets after the count reaches 2-0. So putting the ball in play at 2-0 is a net negative at that point, on average.

Those are the numbers for the league as a whole. What about Brian Bannister in particular? Here are his linear weight run values by count for 2007. The values for a ball, a strike, or a ball in play at a given count are shown relative to the league average overall run value for that count (shown in the previous table).

Since Bannister performed much better than average in run prevention, it’s no surprise that he did better than average in 8 of 12 counts. The counts where he did the best compared to expectations are 0-2 and 2-0. The counts where he did the best overall are 0-2, 1-2, and 0-1. Compared to the average pitcher, he performs much worse when the count is 3-1.

Now that we’ve established a baseline, we can check whether Bannister throws his pitches in favorable counts more often than other pitchers.

Bannister throws a first-pitch strike 4% more often than average. He also pitched more often than average in a 1-1 count and less often than average at 0-2 and 1-0.

We are now in position to evaluate one of his claimed reasons for success: “[H]ow can a Major League pitcher try and beat the .300 BABIP average? By pitching in 0-2, 1-2, & 2-2 counts more often than the historical averages of pitchers in the Major Leagues.” In fact, in 2007, Bannister pitched less often in those counts than the average pitcher, entering the 0-2, 1-2, or 2-2 count with 63% of batters he faced, compared to the league average of 66%.

So maybe he wasn’t accurate about which counts give him an advantage, but perhaps he pitches in advantageous counts overall anyway? As my son would say, “Let’s see!”

It turns out the answer is yes, at least a little bit. Depending on whether you use Bannister’s own performance in each count or the league average performance in each count, you end up with an advantage to Bannister between 1.3 and 2.6 runs based on pitching in favorable counts. Let’s settle on a figure half way in between since I don’t currently have any better idea of how to regress these numbers to the mean. He gives up an extra 2.2 runs by pitching less often than average at an 0-2 count, but more than gains that back by pitching less often at 3-0 (-1.5 runs) and more often at 0-1 (-1.1 runs). He also pitches less often at 3-1 (-0.7 runs) and 2-0 (-0.6 runs).

A two-run difference hardly begins to explain the 15 runs by which Bannister outperformed league average BABIP in 2007. In fact, it may have nothing to do with BABIP at all. We could have guessed that Bannister didn’t pitch in 3-0 or 3-1 counts very often since he allowed less walks than average. In the interview Bannister was mainly talking about balls in play, not pitches rung up for balls or strikes, so let’s get to the heart of the matter and focus on balls in play.

Bannister claimed he’d been able to get batters to put the ball in play 155 times with a 0-2 or 1-2 count versus 78 times at 2-0 or 2-1. In 2007, batters put the ball in play against Bannister 78 times at 0-2 and 1-2 versus 61 times at 2-0 or 2-1. The numbers he quoted were for his career, so to check them, we need to add in his 2006 numbers, 19 and 17 respectively, to get a total of 97 career balls in play at 0-2 or 1-2 and 78 career balls in play at 2-0 or 2-1. The second number matches what Bannister said, but the first number isn’t even close. Did Bannister leave out a count? I can’t find any other combination of counts that adds up to the numbers Bannister said. You can look for yourself.

Inconsistencies in Bannister’s arithmetic aside, the main thing to note is that he did not induce balls in play in advantageous counts more often than average. He comes out at slightly less than a one-run disadvantage overall, including, notably, a 1.6-run disadvantage at 0-2. When looking at his run values for balls in play, we need to remember that small sample size affects the reliability of these numbers. You might be able to chalk most of this effect up to chance. In any case, Bannister is certainly not gaining any advantage here.

Where does that leave us? Are we left knowing nothing more than when we started other than that the reasons Bannister offered for his BABIP performance don’t hold water? In Part 2, we’ll move on to the PITCHf/x data set and see if learning more about his arsenal can help us answer the BABIP question.

Part 1
Part 2
Part 3

What if we knew what type of pitches every major league pitcher threw? What if we had detailed pitch-by-pitch data about how he used those pitches in every game situation? What if this information was accurate and freely accessible to baseball researchers?

Let’s begin with some history. Since Sportvision’s PITCHf/x system was unveiled during the 2006 playoffs, people have been thinking about using the detailed pitch data to classify pitches by type. Reference this comment by MLBAM’s Director of Stats, Cory Schwartz:

“When the system is installed in all 30 ballparks, it will provide unprecedented accuracy, consistency and depth of data to the measurement of speed and trajectory of each pitch,” Schwartz said. “Ultimately we’ll be able to use this data to determine the pitch type in real time and with greater accuracy than ever before. By recording all of this data in real time, we can provide it to broadcasters such as FOX, in-stadium scoreboards, fans via Enhanced Gameday, clubs and other business partners.

It wasn’t long before Baseball Analysts’ Joe Sheehan was leading the public research down that path, too, publishing articles in the spring of 2007 about pitch classification for pitchers like Jeff Weaver, Mike Mussina, and Kenny Rogers, using the data from the 2006 playoffs.

In April 2007, the PITCHf/x system was installed in nine ballparks, and this produced a wealth of data that encouraged more people to join the analysis fun. Dan Fox, Bill Ferris, and Steve West were among the leading PITCHf/x researchers in the first half of 2007, and although the work in the field covered a number of topics, pitch classification was often at the forefront.

Soon the quest turned toward developing a set of rules to classify pitches for many pitchers, perhaps for every major league pitcher. John Walsh published the early definitive article on this topic. In August, the analysis really began to heat up; for example, see these articles from John Beamer and Joe Sheehan. The quest for a pitch classification algorithm was on.

Several people mentioned clustering algorithms as a potential tool for universal pitch classification. K-means clustering was examined, but it requires the user to specify the number of clusters in advance, which limits its usefulness over the whole major league population, where pitchers have widely varying repertoires. Josh Kalk and Joe Sheehan made the most headway with pitch classification by clustering algorithm. Josh Kalk graciously made his results freely available in the former of player cards for every major league pitcher with at least 100 pitches recorded by the system.

Unfortunately, I’m not enough of an expert in clustering algorithms to replicate their work, and neither Joe nor Josh has made their algorithms public. How do we evaluate the effectiveness of their algorithms, or even attempt to construct a different classification method on our own?

I should mention that there is another approach to pitch classification that has been applied quite successfully by John Walsh in answering a number of questions, not least in his article in The Hardball Times Baseball Annual 2008 investigating the cause of platoon splits. John classified pitches into boxes based on speed and horizontal and vertical spin deflection (pfx_z vs. pfx_x, termed “break” by many analysts). This approach appears to work well for large groups but less well for individual pitchers whose fastball speed or pitch movement varies from the norm.

I began investigating the repertoires of individual pitchers with the hope of developing a set of rules I could apply generally to all pitchers. It soon became clear to me that this was not going to be an easy task. Let me recount some of the steps I took, how my results compared to those from Josh Kalk’s algorithm or John Walsh’s approach, and what I learned along the way.

For my first real pitch classification challenge, I chose Boston Red Sox closer Jonathan Papelbon. Try using pitch speed and horizontal and vertical spin deflection to identify those pitches. I don’t think you can do it. Even Josh Kalk’s clustering algorithm fails pretty badly for Papelbon. It’s clear that he throws a fastball, but what type of fastball, and does he throw more than one type? What are his off-speed pitches?

To answer these questions I turned to a paper by Physics of Baseball professor Alan Nathan on calculating spin parameters from PITCHf/x data. I developed a classification technique based on solving some approximate equations Alan presented for spin direction and spin rate. Spin direction and spin rate parameters contain the same information as the vertical and horizontal spin deflection parameters, but they present the information through the lens of a different coordinate system, and sometimes that can be very helpful in differentiating pitches.

In Papelbon’s case, taking the spin view rather than the “break” view is almost essential. Spin direction clearly delineates his curveball and slider, but the difference in their spin rates results in similar overall spin-induced movement that makes them difficult to tell apart on the “break” graph. The four-seam and two-seam fastballs have a similar situation–easily differentiated by pitch speed and spin direction, but the lower spin rate of the two-seamer brings its “break” back close enough to that of the four-seamer that they are hard to tell apart. And finally, while an off-speed pitch, perhaps a changeup, is somewhat identifiable in the speed-break graphs and probably correctly identified by the clustering algorithms, it’s even easier to identify in the speed-spin direction graph. To find its name–the split finger fastball–I turned to scouting reports, which seemed to fit the data. I also found that Papelbon used his splitter mostly against lefties and his breaking ball mostly against righties.

Papelbon turned out to be the classic case where speed-break graphs and clustering algorithms based upon them spit out wrong answers, but spin parameters make his pitches easy to classify. What I had thought was a hard case turned out to be an easy case when the right tools were applied. Other pitchers would not be so easy, and I would find that spin parameters, while continuing to be very illuminating, unfortunately did not lend themselves to a consistent set of rules across all pitchers. Nonetheless, what tentative classification rules did I develop from the Papelbon analysis?

Speed vs. spin direction can be used to identify pitch type clusters that do not appear distinctly in speed vs. break.

• For right-handed pitchers, four-seam fastballs generally have higher speed and a spin axis at a lower angle than two-seam fastballs.
• For right-handed pitchers, curveballs generally have a lower speed and a spin axis at a lower angle than sliders.
• For right-handed pitchers, split-finger fastballs have a lower speed and a spin axis at a higher angle than other fastballs.
• Split-finger fastballs are used primarily against opposite-handed hitters.

Next, I applied the first rule to separate Edinson Volquez‘s four-seam and two-seam fastballs, again something that clustering algorithms based on break have trouble doing (Kalk, West).

For my next challenge, I took on Greg Maddux. Maddux’s pitch repertoire is a tough nut to crack by any method, since by some accounts he purposefully varies the speed and movement on his pitches by slight amounts all the time. The main addition garnered from the Maddux analysis was a first encounter with the cut fastball as a distinct pitch. In retrospect there may have been a few cutters hiding out in the fastball groupings for Papelbon, but they weren’t many in number and they were similar enough in movement to the four-seam fastball that I didn’t detect them as a separate group. With Maddux there was obviously a separate fastball group with slightly lower speed and a spin axis at a lower angle. The speed and spin axis were too low to be regular fastballs and too high to be sliders. In addition, he used the cutter primarily against lefties, whereas sliders and regular fastballs were used more against righties. So we add a couple more classification rules to our list.

• For right-handed pitchers, cut fastballs have a speed roughly 2-4 mph slower and a spin axis at a somewhat lower angle than a regular fastball.
• Cut fastballs are used primarily against opposite-handed hitters.

I then tackled underhand pitcher Chad Bradford, but submariners and true side-armers are such a different breed that I’m going to leave them out of the discussion for now and assume they must be classified by a separate algorithm. Fortunately there are maybe only a dozen of them in the majors, so dealing with them separately or simply excluding them from some studies is not unreasonable.

My analysis of Zack Greinke is less interesting for finding new rules, although it did confirm that speed and spin direction were more useful in separating his slider and curveball than speed and break. However, something at least suggested by Greinke’s case is that a pitcher’s repertoire may look different in relief than in the starting rotation. He may throw harder in the bullpen than in the rotation. In Greinke’s case, he only threw 1-2 mph harder in his relief role, but I suspect the difference may be more for some other pitchers. In addition, a reliever might restrict his repertoire to rely primarily on one or two of his best pitches, while a starter may find more need for third, fourth, or even fifth pitches. This is an avenue for further investigation, but I will list it as a tentative rule.

• It may be useful to analyze a pitcher’s relief outings separate from his outings as a starting pitcher.

Next up is Eric Gagne, for whom the speed/break clustering is actually pretty straightforward. For some reason, Josh Kalk’s player card for Gagne gets some strange results, totally missing his slider. That may be a problem with some of the corrections Josh applies to his data (park, air density), or it may be a problem with his clustering algorithm. I noticed a few other things in the Gagne analysis that are worth a comment. First, many of Gagne’s sliders had a very low calculated spin rate, between 200-700 rpm, compared to 2000+ rpm for a typical fastball. This effect occurs because most of the spin is around the direction of travel. Only the component of spin in the x-z plane affects the break of the pitch, and that is what can be calculated by my analysis. The low measured spin rate is a signature of the slider, and I’ll list that as a new rule here, but we’ll add a qualifier later on.

Also, scouting reports will tell you that Gagne’s off-speed pitch is a “Vulcan” changeup, describing the grip he uses for that pitch. It’s a hybrid of a split-finger fastball and a normal changeup, and it moves like a hybrid pitch. It has the speed of a typical changeup, but the angle of its spin axis is quite a bit greater than that of Gagne’s fastball. Tweener pitches like this are what make comprehensive rule construction difficult. For Gagne we define it as a changeup since its speed is 10 mph slower than his fastball, but will that be universally helpful in distinguishing splitters from changeups? Possibly. Papelbon is a counterexample; however, where his off-speed pitch is about 10 mph slower than his fastball and with a similar spin axis offset as Gagne’s Vulcan change. In terms of absolute speed, Papelbon’s splitter runs 84-91 mph and Gagne’s changeup 80-87 mph. Hmm. Both these rules need to be refined, but we’ll list them anyway.

• Pitches with very low calculated spin rates tend to be sliders since much of the actual rotation is around the direction of travel (and is thus not measured).
• Split-finger fastballs tend to be thrown harder than changeups.

With Josh Beckett, I encountered another important principle and possibly a difficult hurdle for a universal classification system. The speed of Beckett’s fastball, as recorded by the PITCHf/x system, varied by as much as 5-6 mph between starts. It’s possible that was a real effect, or perhaps a data error, but in any case, it made his fastball difficult to distinguish from his changeup. In order to distinguish the pitches, I identified by speed which pitches were fastballs on a start-by-start basis and then normalized the median fastball speed in any given start to Beckett’s overall average fastball speed. Josh Kalk’s park correction seems to be able to overcome this problem in Beckett’s case.

I also observed with Beckett that his four-seam fastball, two-seam fastball, and cut fastball were difficult to distinguish. Two-seamers and cutters have lower spin rates than four-seamers (and with most pitchers, slower speeds, but not for Beckett), but the transition between types was smooth rather than clustered. I didn’t come up with a good solution for this problem, finally just drawing a couple lines through the data where it seemed best to me (where I found minima in the spin direction histogram). I did note that Beckett used his two-seamer more often to right-handed hitters than to lefties, something I also observed with Greg Maddux, but it’s a tendency rather than a hard and fast rule.

• If a pitcher’s fastball speed varies greatly from appearance to appearance, pitches may need to be classified on an appearance basis rather than a season basis.
• Two-seam fastballs tend to be used more often against same-handed hitters.
• Cut fastballs tend to have slower spin rates than four-seam fastballs.
• Two-seam fastballs tend to have slower spin rates than four-seam fastballs.

The next pitcher under my microscope was Joba Chamberlain. His four pitches–fastball, slider, curveball, changeup–were readily identifiable, either by speed/break or speed/spin analysis. (For some reason, Josh Kalk’s algorithm inexplicably identifies all his off-speed pitches as sliders.) Chamberlain didn’t break any of my previously established rules, so while very interesting from a baseball perspective, he’s fairly uninteresting for algorithm development.

Finally we arrive at the start of my tenure with MVN, where I began with an article on John Smoltz. As I did with Josh Beckett, I found it useful to classify pitches for Smoltz on a start-by-start basis because of significant variations by start in spin/break as well as speed. This approach was helpful in distinguishing fastballs from sliders and changeups, or at least that’s how I used it on the first pass. A closer look at the data, combined with reading of scouting reports, informed me that Smoltz had a split-finger fastball as well as a changeup. Apparently at one point in his career the splitter was a primary out pitch for Smoltz, but it seems to have fallen into disfavor in the last couple years because of injuries. Nonetheless, a careful attention to the game-by-game splits of spin direction and speed indicated the presence of a split-finger pitch thrown with slightly greater speed and a greater spin axis angle than the changeup. An exact dividing line between the two was difficult to establish. Given what I learned; however, I can refine my earlier rule from Papelbon and Gagne that simply stated that splitters were thrown harder than changeups.

• Split-finger fastballs tend to be thrown a few mph harder than changeups.
• Split-finger fastballs tend to be thrown with a spin axis at a somewhat higher angle than a changeup.
• Split-finger fastballs tend to be thrown almost exclusively to opposite-handed hitters.
• Changeups also tend to be thrown to opposite-handed hitters, but the tendency is much weaker than for splitters.
• If a pitcher’s speed, spin, and/or break vary considerably from appearance to appearance, pitches may need to be classified on an appearance basis rather than a season basis.

After I published the article on Smoltz, I got some good feedback at the Baseball Think Factory newsblog. It’s apparent to me from that discussion that I’ve lumped the cut fastballs and sliders together in my Smoltz analysis. I can see a handful of 90+ mph pitches which are clearly cutters, but as of yet I have not been able to determine a way to reliably distinguish the two pitches for John Smoltz.

My next article was on Mariano Rivera, who supposedly has a very simple repertoire–one pitch: the cut fastball. In fact, I found that he threw two pitches, mostly the cutter but also a regular fastball, subsequently confirmed by searches of the scouting reports. This fact is visible but overlookable in the speed/break graphs, and in fact Josh Kalk’s clustering algorithm only sees the cutter. Looking at spin direction, however, two clusters are readily identifiable, and this identification is further confirmed when one notes that Rivera only throws the regular fastball to righties. As an aside, I went the extra mile on identifying the pitches at the cutter/regular fastball boundary by examining the spin direction on an appearance-by-appearance basis. This is probably not strictly necessary for a universal algorithm that is allowed to misdiagnose a pitch here or there, but for getting closer to 100% accuracy it was useful in this case.

My most recent pitcher analysis was for Joakim Soria. Finally we find another pitcher for whom Josh Kalk’s algorithm and I are in pretty good agreement. One area of difference is in what to call his fastball. It has a fair amount of cutting movement, and his catcher called it a cutter, so I called it a cutter, but I wouldn’t argue too strenuously with someone who wanted to call it a regular four-seamer. It’s on the borderline, and one facet of designing any universal algorithm is going to be a decision on what to call pitches like Soria’s fastball. One interesting thing to note for Soria is that his slider looks like a curveball from many other pitchers, in its speed, break, and spin direction. What identifies it as a slider for Soria is comparison to his curveball, which is slower, breaks more, and has a spin axis at a smaller angle. Soria brings up more questions than new rules, but we’ll list them anyway…

• Should fastballs without much lateral movement that are used as a pitcher’s primary fastball be labeled as cut fastballs or four-seam fastballs?
• Should sliders and curveballs be identified primarily by comparison within the pitcher’s own repertoire or with respect to the sliders and curveballs of the rest of the league?

With Soria I also observed that he throws his slider almost exclusively to righties and his changeup almost exclusively to lefties. Not every pitcher is nearly as strict in their division of work between the two pitches, but it’s a general trend across pitchers.

• Sliders tend to be thrown more often to same-handed hitters.
• Changeups tend to be thrown more often to opposite-handed hitters.

Expanding on the idea we touched on with Soria, a slider can be a very different animal from one pitcher to another. Some sliders look almost like cutters (John Smoltz), some look almost like curveballs (Joakim Soria, Francisco Rodriguez), and others have only a little spin-induced movement (Brad Lidge, Jeremy Bonderman, Felix Hernandez). Yet all of them are called sliders, and all (except for Soria) are well-known as among the best sliders in the game. I’ve been attempting to gain a better understanding of the slider in order to get a better handle on how to classify a pitch with so many sub-species.

One thing I’ve learned, thanks to John Walsh, is that sliders without much spin-induced movement–such as those we identified earlier in the Gagne analysis–have, in addition to a low x-z component of the spin rate, a great deal of error in the calculation of the direction of the spin axis. Such sliders have a spin axis pointing nearly in the direction of travel (the y direction), and any x-z component of the spin axis is small and highly susceptible to measurement error.

• Spin direction calculations have a high degree of error for sliders with very small spin-induced movement.

Another pitch that deserves a quick mention is the gyroball. What is it? I know of one pitcher who definitely throws the gyroball: C.J. Wilson. I even have identification by Wilson from his own blog of a few specific game instances when he threw the gyroball. From my previous reading, I had expected the gyroball to be slider-like, but Wilson’s gyroballs look like fastballs, so much like fastballs that I have been unable to identify them as an otherwise separate pitch from his fastball. This is an area for further investigation.

I’ve also been investigating a few new ideas in pitch classification, including Tom Tango’s suggestion of “reaction break” and adjustments to the spin-induced movement and spin rate calculations based on new formulas from Alan Nathan. Both of these show promise, but I’m still in the early stages of investigating their application.

In summary (at last, you say?), we’ve come a long way in the science of pitch classification since spring 2007. We have the invaluable reference of Josh Kalk’s player cards. We have a sophisticated set of tools for classifying the repertoire of an individual pitcher. However, we are still not near the point of having a universal system for classifying >95% of major league pitches accurately. A published system with that level of accuracy would serve as the spark to ignite a great deal of other research that could transform the game of baseball.

Perhaps Sportvision and MLBAM are developing such a system in private. Perhaps Joe Sheehan’s unpublished system is closer to that goal than I realize. Perhaps Josh Kalk is on the way to making refinements to his system that will get us closer. Maybe John Walsh’s approach, which went to press as I was writing this article, of using K-means clustering and going through the major leagues pitcher by pitcher is the right way to start.

Clearly, one of my priorities should be improving my understanding of and facility with clustering techniques. In my opinion, an effective clustering algorithm should include speed, spin direction, spin rate, vertical spin deflection, horizontal spin deflection, and handedness of the pitcher and hitter. It’s possible that additional variables such as strike zone location, release point, and ball-strike count might be helpful, but I have a feeling their addition to a clustering algorithm might cause more problems than they solve.

I would like to develop a classification system that would be available to any researcher and accurate enough to distinguish splitters from changeups, sinkers from four-seam fastballs, and cutters from sliders and fastballs. I confess to being a bit daunted by the task and not very experienced in some of the statistical math that may be required, but I’m going to keep chipping away at this problem until we get there.

Recent evidence may suggest otherwise, but I am still a contributor to Statistically Speaking. I’ve been working on an analysis that has been more difficult to bring to fruition than I expected; that, along with “real life” getting in the way more of late, is what has severely cut into my posting frequency.

However, in the process of number crunching for the analysis I’m doing, I came across some statistics that I haven’t seen posted publicly anywhere, not even in the Baseball-Reference splits. (Some of it is in the B-R splits, but not most of it.) Maybe I’ve just missed them, in which case drop me a line and let me know where else you found them. I thought these might be interesting to a few other people, so I’ll share them. Mostly, I’m just putting the numbers up here for the rest of you to enjoy, but I’ll also make a few comments on some trends that stuck out to me.

I’m looking at pitch data broken down by ball-strike count. I’m using the MLB Gameday 2007 data as my source. Today I present the breakdown of types of balls put into play by the hitter.

A hitter reaches base safely more often on balls in play when the count is in his favor. Don’t change the channel, the revelations like that just keep on coming at StatSpeak, and you don’t want to miss one!

Okay. My first slightly less than completely and utterly obvious observation is that the home run rate is strongly tied to the count.

The doubles rate shows the same effect, but smaller, as does the triples rate to some extent. The singles rate stays pretty flat with respect to count, although there is a bit of an inverse effect–in better hitter’s counts, the hitter gets more extra base hits and slightly fewer singles.I haven’t looked at the type of batted ball (fly ball, line drive, ground ball, bunt, etc.) that results in hits. That’s a bit more difficult to parse out of the Gameday data. Since it doesn’t have its own field, getting that information requires some regular expression matching on the text description of the play. That’s fairly straightforward but nonetheless a nontrivial bit of coding that makes it a project for some point in the future rather than part of this data set for me.

Another thing I noticed was that there were more groundouts and less flyouts the more strikes and less balls there were in the count. As pitchers gain the upper hand, they tend to get more groundball outs. I didn’t include popups and line drives in the accompanying chart since they didn’t show a strong tendency relative to count.

I saw a couple other things that are obvious once you think about them, but it was interesting to me to see them reflected in the data. The first was that force outs, GIDPs, and fielder’s choice outs all go down dramatically with a 3-2 count, dropping from 6.4% to 2.3% of balls in play. Presumably this is because the runners are often going with the pitch on 3-2.

The second thing that interested me was the favorite counts for hitters to bunt for an out. (Bunting for a hit is not included for the reason mentioned previously.)

If I don’t get around to presenting my full analysis in a timely fashion, I’ll see if I can present a few more statistical tidbits like this along the way.

I don’t know any other major league pitcher who relies on his cut fastball to nearly the same extent as Mariano Rivera, but there are many pitchers who use a cutter to some degree. Most of them, like Josh Beckett, merely put a little “cut” on a fastball now and then, and it’s debatable whether to classify it as a separate pitch in their repertoire. Some of them, like Greg Maddux, throw both a cut fastball and another fastball as fairly distinct pitches. A few others, like our subject today, throw a single type of fastball that moves more like a cutter than it does like a traditional four-seamer. Do we also label this kind of a pitch a cut fastball?

The cutter is second only, perhaps, to the slider in the flexibility of its definition. Almost every starting pitcher is said to throw a cutter by an obscure report somewhere. I’ve learned to discount these notional references, but I pay a lot more attention when the pitcher himself or his catcher says he threw a cutter.

Which brings us to Joakim Soria, closer for the Kansas City Royals. The Royals picked him up from the San Diego Padres in the Rule 5 draft last winter, and what a find that was! He had been pitching well in the Mexican League, and showed his stuff for the Royals last year when the closer of plan, Octavio Dotel, was first injured and later traded. Soria appeared in 62 games, pitched 69 innings, allowing 46 hits, 19 walks, and only three home runs, while racking up 75 strikeouts to go with 17 saves and 2.48 ERA.

What pitches does Joakim Soria throw? His catcher John Buck reports:

“It’s hard to pick him up. His ball has a natural cut to it. Not as much as [Rafael] Soriano but it does have a cut to it. That’s just his natural fastball,” Buck said.

“He has a great slider and curveball and can throw his change-up on any count. You have to kind of speed up your bat to get the head up to hit the cutter and, all of a sudden, he throws a changeup and it makes it difficult — sitting in-between those two is a tough place to be as a hitter.”

So his catcher calls his fastball a cutter. Let’s take a look at the data we have from PITCHf/x for the 2007 season, covering 477 pitches for Joakim Soria. I’ll begin with a graph of pitch speed versus the angle at which the spin on the ball is deflecting the pitch.

Soria has a fastball with a lot of cut that runs 89-94 mph. The cut fastball is his bread-and-butter pitch; he uses it for 69% of his pitches to lefties and 78% of his pitches to righties.

He has a changeup with a lot of lateral action that he throws 80-84 mph. He uses the changeup almost exclusively to lefties, making up 19% of his pitches to them.

As his off-speed pitch to righties, Soria uses a slider with a big break that runs 76-81 mph. The slider makes up 11% of his pitches to right-handed hitters.

Rounding out his repertoire is a slow curveball that Soria throws 66-71 mph. The curveball makes up 10% of his pitches, and he uses it equally to lefties and righties.

Let’s look at how these pitches move from the hitter’s perspective.

All of Soria’s pitches have good movement. His fastball has”cut” to it, and his changeup has good lateral and vertical movement when compared to his fastball. His slider looks like most pitchers’ curveballs, and his curveball is a slow ball with a lot of drop.

Next, let’s look at what pitches Soria throws in each ball-strike count.

And here’s the same information presented graphically:

We can see that until he gets a strike, Soria uses almost only the cut fastball, and when he gets two strikes, he brings out the curveball pretty often, except in a 3-2 count, where he sticks with the cutter. This would imply that the curveball is his strikeout pitch and that he has trouble getting strikes with his off-speed pitches.

As a second opinion, you can look at what Josh Kalk’s algorithm spit out for Joakim Soria. Josh also has release point data there if you are interested in that.

Finally, let’s examine where Soria throws his pitches and what results he gets.

–-
CS=called strike, SS=swinging strike, IPO=in play (out), IPNO=in play (no out), TB=total bases, BABIP=batting average on balls in play (including home runs), SLGBIP=slugging average on balls in play (including home runs). For Strk% all pitches other than balls are counted as strikes. Con% = (Foul+IPO+IPNO)/(Foul+IPO+IPNO+SS).

Our earlier conclusions seem to hold up.

Here are Soria’s results for the cut fastball.

To lefties, Soria seems willing to pound the zone with the cutter, and his results indicate that strategy works. Against righties, he works more up and away. He misses the zone a little more often, and he generates more foul balls, but his results are still good.

Moving on, let’s see the results for the changeup and slider:

As I mentioned earlier, Soria uses the changeup to lefties and the slider to righties. In both cases, he likes to throw down and away. It looks like he has trouble throwing the slider consistently for strikes.

Last, but not least, the curveball.

Soria gets a lot of swinging strikes in the zone to both lefties and righties. The only difference appears to be when he misses–down and away to righties, and up and away or down and in to lefties.

Since I mentioned earlier that the curveball looked like Soria’s strikeout pitch, let’s check on that. We have PITCHf/x data for 40 of his 75 strikeouts. For those 40 K’s, 23 of them were on the curveball, 9 on the cutter, 4 on the changeup, and 3 on the slider.

I hope you enjoyed the analysis of one of my favorite players from my favorite team. My work’s had a bit of an “East Coast bias” lately, which feels a bit odd to me. I don’t expect to continue solely in that vein. If nothing else, you should see a Royal popping up in this space now and then.

Despite winning the American League West with a 94-68 record last year, the LA of Anaheim Angels have gotten short shrift from the PITCHf/x analysts thus far. The only writeup that the pitching staff has gotten was one by Joe Sheehan on John Lackey three weeks into the season. I’d like to remedy that a little bit today. The Angels had three outstanding starters: Lackey, Kelvim Escobar, and Jered Weaver. Let’s take a detailed look into the pitching performance of Kelvim Escobar.

Escobar is a 31-year-old right hander from LaGuaira, Venezuela. He was a former starter turned reliever (and closer) and back to starter again for the Toronto Blue Jays before joining the Anaheim Angels in 2004. He’s struggled to stay completely healthy, but overall he has turned in some fine numbers for the Angels in four years: a 43-35 record and 3.60 ERA in 109 starts, allowing 611 hits and 213 walks against 561 strikeouts in 653 innings.

Since the Big A was one of the original nine stadiums to have a camera system installed from the beginning of the 2007 season, the large majority of Escobar’s season was recorded by the PITCHf/x system, 2469 of his total 3141 pitches. This gives us a good data set to identify his pitches and examine his pitching tendencies.

Escobar throws quite an array of pitches: a four-seam and two-seam fastball, a changeup and split-finger, a slider and a curveball. According to scouting reports, he is capable with all six pitches.

Here I’ve shown two graphs that I use for pitch classification. The first graph shows the speed of his pitches versus the direction they break, in polar graph format. The second graph shows the movement due to the forces of spin deflection and gravity on his pitches in the last quarter-second before they cross the plate.

There are a couple other ways to look at the vertical vs. horizontal deflection over the whole pitch trajectory:

Escobar’s four-seam fastball runs 92-96 mph, and the average spin deflection he gets on the four-seamer is a 10-inch hop and a 4-inch tail in toward right-handers. Compared to a league-average fastball, that’s 3 mph faster but with a couple inches less lateral movement, probably due to the fact that his motion is more over-the-top than many right-handed pitchers. The four-seamer is one of Escobar’s main pitches to both lefties (26% of the time) and righties (27%).

Escobar’s two-seam fastball also runs 92-96 mph, but its average spin deflection is an 8-inch hop and a 7-inch tail in toward right-handers. The two-seamer is his primary pitch to lefties (28% of the time) and also a main pitch to right-handers (24%). I made the division between the four-seamer and the two-seamer by looking at the spin direction of each pitch on a game-by-game basis, but the dividing line between the two is still a bit fuzzy to me.

His split-finger fastball runs 85-89 mph, and its average spin deflection is a 6-inch hop and a 6-inch tail in toward right-handers. Escobar uses the splitter fairly often to left-handers (15% of the time) but only infrequently to right-handers (6%).

His changeup runs 83-87 mph, and its average spin deflection is a 10-inch hop and a 3-inch tail in toward righties. The 9-mph separation between his fastball and changeup is about average for major league pitchers. He uses the changeup more often to lefties (16% of the time) but also some against righties (11%).

Escobar’s slider runs 85-89 mph, and its average spin deflection is a 3-inch hop and a 2-inch break away from righties. That’s about 3 mph harder than the average major-league slider, with typical movement. The slider is one of his favorite pitches to right-handed hitters (25% of the time) and is rarely used against lefties (2%).

Finally, his curveball runs 79-84 mph, and its average spin deflection is a 3-inch drop and a 1-inch break away from right-handers. That’s about 4 mph harder than the average major-league curveball, with 12-to-6 movement that is somewhat rare. (The spin deflection on the average major-league curveball is a 2-inch drop and a 5-inch cut. John Walsh’s article is my source for league average numbers.)

Next, let’s look at how Escobar mixes his pitches in different ball-strike counts, which I’ve split out by batter handedness. The picture gets a bit messy when a man throws six different pitches, but let’s dive in and see what we see.

To lefties, Escobar uses the four-seamer on any count and relies on it a little more if he falls behind. He throws the curveball early in the count, 22% of the time with no balls, 9% of the time with 1 ball, and only 3% of the time with 2 or 3 balls in the count. He favors the two-seamer with 0 or 1 strike, 33% of the time, but uses it only 16% of the time with 2 strikes. Instead, with 2 strikes he relies on the splitter 32% of the time. He’ll throw the changeup at almost any count except 0-2 and 3-0, but he likes to throw it more when he’s behind in the count, in which case he throws it 25% of the time.

Early in the count with Escobar, lefties should expect to see the two-seamer, the four-seamer, the curveball, and the changeup, in that order. If Escobar gets the hitter down 0-2 or 1-2, he should expect the splitter (41% of the time) or perhaps a fastball (41%), but if the count goes 2-2 or 3-2, he should start to watch for the changeup, too (33%).

To righties, early in the count, Escobar throws hard stuff, 31% two-seamers, 28% sliders, 23% four-seamers, and only 18% of his other three pitches combined. When he gets 2 strikes, the two-seamer disappears (only 3%), but he’s willing to show the splitter (14%). The changeup gets used a little with 1 strike (11%), but at 2-1 or 2-2 it’s a favored pitch (26%), and at 3-2, it’s his favorite pitch (34%), like it was to lefties. Righties can expect the curveball mainly at a single count: 0-2, where Escobar uses it 28% of the time; it’s little used (6%) in other counts.

What kind of results does Escobar get with each of his pitches? His four-seam fastball is a pretty good pitch, but his two-seamer grades out poorer. All four of his off-speed pitches are above average. I should mention that the PITCHf/x games for Escobar are missing his two worst starts of the year, which skews all the following numbers a little bit in his favor.

The league average information comes from John Walsh’s article. In the following pitch location charts, I’ve changed my color-coding a bit to try to improve readability for those with color blindness. Hopefully the new system is an improvement.

Escobar works with the four-seamer on the outer half of the plate to both lefties and righties, although with lefties he works down more and avoids coming inside, and with righties he works up more and works inside just off the plate. He has some trouble throwing the four-seamer for strikes to righties (only 59%, compared to 64% league average), but when he does, and they put in play, he gets very good results: .176/.216 (avg/slg), compared to .330/.521 major-league average off the fastball.

To lefties, he’s much better at throwing the four-seamer for strikes (67%), and he gets a lot of called strikes (26% compared to 19% league average), but his results on balls in play are only fair: .315/.444 avg/slg. He didn’t allow a single home run in 31 fly balls hit off the four-seamer in PITCHf/x games. That is unusual–fastballs are the most homered-upon pitch for most pitchers.

Escobar has trouble throwing the two-seamer for strikes, getting it over only 60% of the time. As with the four-seamer, he works mainly on the outer part of the plate to both lefties and righties. However, both lefties and righties have good success when they put the two-seamer into play. Lefties hit .338/.523, and righties hit .415/.585.

The splitter is Escobar’s strikeout pitch to lefties, and you can see why. They swing and miss at it down and away more often than not. He doesn’t necessarily throw it in the strike zone that much, but he gets strikes because the hitters chase it. When he does get it in the zone, hitters do much better with it, making at least decent contact and racking up a .294/.441 line, including a home run.

He doesn’t throw the splitter nearly as much to righties, although I wonder if maybe he should. He still gets a lot of swings and misses (19%, compared to 13% league average), but righties are able to put the ball in play almost every time he gets the splitter in the zone. However, the right-handed hitters don’t fare nearly as well as lefties on balls in play, hitting only a meager .158/.211. Perhaps it’s the small sample size (19 balls in play), or maybe righties really do have trouble getting good wood on the splitter.

The changeup is the first pitch where we see a marked contrast in Escobar’s location to lefties and righties. To lefties, he pitches away, away, away. He gets some swings and misses in the zone, but lefties don’t chase the changeup out of the strike zone much. On balls hit into play by lefties, Escobar does well, a .216/.297 line, compared to .319/.502 against an average major-league changeup.

To righties, he throws the changeup mostly in the zone or on the corner low and away. He gets a lot of swings and misses, especially on the outside corner. The changeup is a very effective pitch against righties. No wonder he likes to throw it as a strikeout pitch to righties. Moreover, even though he pounds the heart of the zone, righties have little luck on balls in play, hitting only .237/.316. Most right-handed pitchers avoid throwing the changeup to right-handed hitters, but for Escobar in that situation, it’s a great pitch and one he could perhaps use even more often.

As you can see, his slider is rarely used to lefties, mostly thrown up and in and fouled off. To righties, he uses the slider a lot, and to good effect. He gets a good number of called strikes (17%, versus 14% league average) and swinging strikes (16%, versus 13% average), and when the ball is put in play, Escobar also fares well, allowing a .254/.352 avg/slg, compared to .310/.481 against an average major-league slider. Those numbers include allowing only 1 home run on 27 fly balls hit by righties off the slider–luck or skill?

Finally, we come to the curveball, Escobar’s least-used pitch. He throws it mostly down and away to both righties and lefties, although he also throws it in the zone quite a bit. He gets a lot of called strikes, especially to lefties (32%), but also to righties (23%), compared to league average of 19% with the curve. Most pitchers rarely throw the curveball as the first pitch to a batter. Escobar, on the other hand, often throws a lefty a curveball right across the plate for strike one looking. Lefties don’t often make contact with the curveball, but when they do, the results are decent: .353/.412, compared to league average against the curve of .310/.471.

Right-handers see the curveball more often with two strikes, and it’s a good strikeout pitch for Escobar, both swinging (at balls in the dirt) and looking. Righties don’t make contact with the curve very often, either, and when they do, their results are particularly poor: in 13 curveballs in play, righties hit 10 groundballs (including two double plays), 2 fly balls, and one line drive. The line drive and one groundball landed as singles, for a .154 average.

In summary, Escobar has a solid four-seam fastball which he complements with a weaker two-seamer, and his array of off-speed pitches is impressive. His changeup, splitter, curveball, and slider are all well above average pitches, and some of them, particularly his changeup, are among the best in baseball. He struggles with control on his fastball, and this, along with the recurrent health problems, is probably all that keeps him from being one of the very best pitchers in baseball.

As a final note, I thought this was a great photo from MLB.com of Kelvim Escobar in full stride.

–

If you enjoyed this article, you might be interested in my similar previous analysis of Erik Bedard, Johan Santana, James Shields, Mariano Rivera, Joakim Soria, Josh Beckett, Joba Chamberlain, or Eric Gagne.

Who is the best pitcher in baseball right now? Some might answer that question with Jake Peavy or Josh Beckett, but I’d guess that at least 7 out of 10 times, the answer you would get is Minnesota Twins left-hander Johan Santana. Santana is a 28-year-old from Tovar, Venezuela, and after his fourth full year in the starting rotation, he already owns two Cy Young Award trophies.

Now, as Santana approaches the final season of the 4-year, $39.75 million contract he signed three years ago, the Twins appear eager to trade him, and the reported suitors include such teams as the New York Yankees, Boston Red Sox, and New York Mets, subject to Santana’s approval. I’ll leave the predictions of where he’ll land to those who are better qualified or more eager to comment than I am. However, I’d like to take a look at the pitching repertoire and strategy of possibly the best pitcher in baseball.

If you look at the scouting reports, they all talk about Johan Santana’s devastating changeup and how he works to make his throwing motion identical for all pitches. Most scouting reports list three pitches for Santana–fastball, changeup, and slider–and mention that his changeup comes in 15-20 mph slower than his fastball. Were this true, it would be highly unusual. Most major league changeups are 7-10 mph slower than the pitcher’s fastball. A few scouting reports speak of five pitches–two fastballs, a slider, a circle change, and a straight change. The most useful and interesting scouting information I found was an interview from 2006 that Pat Borzi conducted for the Sporting News with Johan Santana and his catcher Joe Mauer.

Santana throws four pitches for strikes-four- and two-seam fastballs between 92 and 95 mph, a slider/curve in the 84- to 87-mph range and a changeup that’s about 15 to 20 mph slower than the fastball. The changeup is his strikeout pitch; when Santana is on, he throws it from the same arm angle and release point as his fastball, and hitters can’t tell the difference until it’s too late.

I also found this quote from Santana interesting given that most people acknowledge his changeup as his best pitch:

“I want to make sure my two-seam fastball is working,” Santana says. “That’s my best pitch, and it’s going to make my other pitches look even better. That’s what I try to do all the time.”

We have detailed data from the PITCHf/x system for 1032 of Santana’s 3345 pitches during the 2007 season. Let’s dive in and see what we can learn about Santana’s repertoire and effectiveness with his various pitches.

Santana has at least three obvious pitch groupings: fastball, changeup, and breaking ball. Here I’ve shown two graphs that I use for pitch classification. The first graph shows the speed of his pitches versus the direction they break, in polar graph format. The second graph shows the movement on his pitches in the last quarter-second before they cross the plate, due to the forces of spin deflection and gravity.

The fastballs run 89-95 mph, and it’s hard to tell from these graphs alone whether Santana really does throw two different fastballs or just one. Through additional analysis, which I will explain shortly, as well as Santana’s own comments, I concluded that he did in fact throw a four-seam and a two-seam fastball and have coded them separately in these graphs.

We can also see that Santana throws two different offspeed pitches. One has a movement very similar to the fastball but is thrown slower at 80-84 mph. This is his changeup. It’s interesting to note that we see a 10 mph difference in speeds between his fastball and his changeup, typical of other major league changeups and nothing like the 15-20 mph difference that was reported by other sources. I don’t know if that was just the stuff of legend or whether Santana has changed his approach in recent years. More likely, people were comparing Santana’s very slowest changeup with his very fastest fastball and writing as if that represented a typical pitching pattern.

I could not find any sign of two different changeups in Santana’s repertoire, at least not two changeups that consistently have different movement or speed.

Santana’s other offspeed pitch is an 83-88 mph breaking ball, described in various scouting reports as either a slider or a curveball. Based on the spin direction, the speed, and the direction of break, it’s very clearly a slider. In the first graph of pitch speed vs. spin deflection angle, the calculation of the spin deflection angle for some of the sliders contains a good deal of error since the spin of those sliders is nearly aligned around the direction of travel of the pitch, resulting in spin deflection of only a couple inches or less. This is one of the classic indicators of a slider.

The sliders and changeups look difficult to separate at the margins in the two graphs I presented above, but including the (x-z component of the) spin rate in the discussion makes that task much easier.

Returning to the topic I mentioned earlier, how did I determine whether Santana threw both a four-seam and a two-seam fastball? Looking at the data in aggregate, it was impossible to see a dividing line, but when I examined the spin and break on a start-by-start basis, a little bit of order appeared out of the murkiness. In some starts, two separate groupings were obvious. In most starts, the dividing line was subtle. In a few cases, it was hard to find a dividing line at all. I did notice that the fastballs with the most sink and the slowest speed were thrown almost exclusively to right-handed hitters, and this, in addition to Santana’s own comments about throwing a two-seamer, gave me confidence in making a distinction between the two fastballs.

If you look at the comments from John Walsh and John Beamer on my Erik Bedard analysis, you’ll see that having to examine the data on a start-by-start basis in order to make an accurate pitch classification diagnosis is a recurring problem. We’d like to be able to look at a pitcher’s season data as a whole. This is an important area for further investigation.

Here are a couple more traditionally-used PITCHf/x graphs of pitch movement for those who are interested:

How does Santana use his pitches to left-handed and right-handed hitters? As a left-handed pitcher, he naturally sees predominantly right-handed hitters, making up 75% of his opponents. To righties, he throws about 41% four-seam fastballs, 35% changeups, 18% two-seam fastballs, and 6% sliders. To lefties, he throws 60% fastballs, 29% sliders, 7% changeups, and 4% two-seam fastballs. Against righties he’s the stereotypical fastball-changeup Santana that I’ve heard about. Against lefties, he’s a totally different pitcher, eschewing the changeup and the two-seam fastball and relying on a fastball-slider combination.

Next, let’s look at how Santana mixes his pitches in different ball-strike counts. I’ve split this out by batter handedness as well.

Against righties, you can see that the changeup is his favorite pitch with two strikes (57% of the time), and he mixes in his two-seam fastball more if he falls behind in the count (28% when behind vs. 15% when ahead or even).

Against lefties, he’s relies on the four-seamer about 70% of the time in most situations. With two strikes he feels confident enough to occasionally (14%) introduce the changeup to lefties, and on an 0-2 count, you can count on getting a slider two thirds of the time.

What’s the bottom line–what results does Santana get with his pitches? I attempted for a while to cast the answer to that question in terms of run values for each pitch determined by linear weights, but I’ve postponed that endeavor for the moment. There are too many pieces that I haven’t figured out how to put together yet. So here are the results in the same format I used in the Bedard article.

The league average information comes from John Walsh’s article, and once again I’m using an adaptation of his format to present this information.

The four-seamer is Santana’s bread and butter, especially to lefties, and a good bit of creamy butter it has. He throws it for strikes and gets more swings and misses with it than most pitchers do. Hitters have a hard time putting the four-seamer into play, and when they do, Santana also gets really good results (a .188 BABIP compared to .304 league average BABIP on the fastball), although lefty batters–Hafner, Sizemore, and Thome–did hit three home runs off the four-seamer in our data set. He mostly pounds the zone with the pitch to both lefties and righties, although there appears to be some tendency toward pitching up and away from lefties and up and in to righties.

Santana doesn’t use the two-seamer much against lefties, and when he did, it was mostly for a ball. He works in the zone against righties and gets fairly average results with the two-seam fastball. One surprising thing to note is that he still gives up a lot of fly balls off the two-seamer; almost 70% of balls in play off the two-seamer were fly balls. The two-seamer seems like his weakest pitch based on the results we have from 2007, so I’m not sure I understand his statement from the Sporting News interview that it’s his best pitch.

Just look at all the red bleeding over the graph from the swinging strikes, and you know all you need to know about Santana’s changeup. The hitters can’t hit it. Santana can throw it for strikes just as well as his fastball. He throws it down and away from righties, and he gets a lot of swings and misses when they chase the changeup down out of the strike zone. When he gets it too close to the heart of the zone, they do make decent contact. It would go without saying, but this is an outstanding pitch.

Against lefties, Santana uses the slider mostly down and away, and he gets pretty average results with it. Against righties, he features the slider less often. When he does throw it, he keeps it inside. When he gets it up, it gets put in play, but he had fairly good results on a limited sample of balls in play except for one slider that Alex Rios launched 414 feet into the left field seats at the stadium formerly known as SkyDome.

I also looked a bit at pitch sequencing. Here’s a table showing what pitch a hitter is most likely to see from Santana based on what the previous pitch was.

I don’t notice any particular patterns to lefties, but to righties he’s more likely to throw the two-seamer after a previous two-seamer, and he’s more likely to throw a changeup after another changeup.

Johan Santana had yet another great season in 2007. He allowed a few more walks and home runs than in previous years, but without PITCHf/x data from previous seasons, I don’t have any way to know whether that was simply luck or a change in his pitching abilities and strategies.

I looked at the 11 home-run balls off Santana for which we have PITCHf/x data, and I couldn’t detect any useful patterns. They were mostly hit off pitches up and over the plate, but that doesn’t come as much of a surprise. Looking at the HitTracker data, he wasn’t burned by many short home runs barely sneaking over the fence, so he wasn’t unlucky in that regard, at least. This may be a topic for further investigation or possibly just the result of Santana being a fly ball pitcher and getting a little unlucky with how hard the hitters hit 33 of those fly balls in 2007.

Santana obviously has an outstanding changeup and a strong fastball, but you probably knew that already. What I didn’t know was how infrequently he uses the changeup against lefties or most of the other nuances of his pitching strategy. Unless you’re Joe Mauer or Mike Redmond (in which case, Hi!), hopefully you feel like you know the best pitcher in baseball a little better than you did before.

If you’re an employee of a Mr. Steinbrenner or a Mr. Henry gathering information for a future trade, by all means feel free to contact to me regarding where to send that check for my services.

Baltimore Orioles ace Erik Bedard had a breakout season in 2007 before being sidelined at the end of August with an injury to his oblique muscle. Prior to his final start, he was receiving strong Cy Young Award consideration with a 13-4 record, a 2.97 ERA, and a league-leading 218 strikeouts and only 135 hits and 52 walks allowed in 176 innings. (His last start, pitching injured, resulted in final season numbers a little worse than these I’ve listed.) Compare his 2007 numbers to his previous career-best season in 2006, when he finished 15-11, with a 3.76 ERA and 171 strikeouts against 196 hits and 69 walks in 194 1/3 innings, and it’s clear he stepped up his game in 2007. How did he do it?

Unfortunately we don’t have any PITCHf/x data from the 2006 regular season, but we can use the PITCHf/x microscope to take a look at what Bedard did in 2007 (in the 701 pitches for which we have detailed data). What does Erik Bedard throw? Let’s take a look at his repertoire by graphing the speed of his pitches versus the direction they break.

Bedard has four pitches, as best I can tell. His famous erstwhile pitching coach will tell you he throws four different types of fastballs, but I can only see a four-seamer and a cutter. Either he throws the sinker and the “comebacker” infrequently, if at all, in game action, or they move too similarly to the four-seamer for me to differentiate them using this data.

I mentioned already that Bedard pitched with an injured oblique muscle in his final start of the year on August 26. Most of the fastballs and cutters with a speed below 90 mph were recorded in that start. He averages just over 93 mph on his four-seam fastball and 92 mph on his cut fastball. In his August 26 start, those clocked at 89 and 88 mph, respectively.

When healthy, his four-seam fastball runs 92-95 mph and breaks away from a right-hander by about 7-11 inches. The four-seamer is one his two primary pitches to right-handed hitters; he throws it 34% of the time. Against lefties, it’s his third pitch, used only 23% of the time.

His cut fastball runs 90-94 mph and breaks away from a right-hander by about 2-6 inches. The cutter is his primary pitch to lefties, used almost half the time (45%); against righties, it’s his third pitch, used 24% of the time.

Bedard also throws an occasional 80-83 mph changeup, almost exclusively to lefties (7%). Probably his best pitch is a 76-80 mph curveball, which he uses equally to righties (35%) and to lefties (32%).

Let’s take a look at how all his pitches move, including the effect of gravity in addition to spin-induced deflection.

The slower pitches like the curveball and changeup drop more because gravity has longer to act on them.

I thought it might also be interesting to show something more in line with what I believe the hitter perceives as the “late break” on a pitch, the deflection of the pitch due to both spin and gravity in the last quarter-second before it crosses the plate. Thanks go to Tom Tango for this idea.

This seems to give a more realistic guess at how a hitter might perceive the drop on a curveball compared to a fastball.

Let’s take a look at how Erik Bedard mixes his pitches in different ball-strike counts.

We can see that he uses his cutter more often early in the count or when he falls behind and his four-seam fastball more often with two strikes. He uses his curveball equally across almost all counts, except for avoiding it on 3-0 and 3-1 and showing some preference for it on 2-1 and 3-2 counts. His changeup shows up mostly on 0-1 and 1-1 counts; he throws it to righthanders 20% of the time on those two counts and only 3% of the time on other counts.

Here’s a table showing the details by count.

Now, let’s examine where in the zone Bedard throws his pitches and what results he gets with them.

The league average information comes from John Walsh’s article, and I’ve adapted his format in presenting this information. His pitch types probably don’t correspond exactly to mine since he lumps sinkers and cutters in with four-seam fastballs and splitters in with changeups. I believe it’s still helpful to use his league-wide information for comparison since I haven’t established a league-wide baseline on my own yet.

With the four-seam fastball, Bedard mostly works the outside part of the plate, especially to lefties but also to righties. To lefties, he mostly stays up or away with the fastball, out of the strike zone, and when he does get in the zone, he doesn’t have very good results, although the sample size is small. Against righties, he gets very good results with the fastball, holding them to a .233 BABIP and a .419 slugging percentage.

With the cut fastball, Bedard works away from lefties and gets a lot of called strikes and not much good contact, although two cutters in the middle of the zone did go for home runs. Against righties, he’s all around the zone with the cut fastball, and his results aren’t quite as outstanding. He gets a few more foul balls and a lot less swinging strikes, but overall his results with the cutter are still pretty good against righties.

Erik Bedard threw one changeup to a lefty, Lyle Overbay, out of the 701 pitches in our data set, and that resulted in a fly ball out. To righties he works the changeup down and away, mostly out of the strike zone. When he gets it in the zone, they make contact. The changeup looks like Bedard’s weakest pitch.

Bedard throws the curveball down and away to lefties, and he generates a lot of swings with it–foul balls, swings and misses, and balls in play. Against righties he also works down and away but isn’t afraid to throw it in the zone. He gets a lot of swings and misses and when the ball is put in play, it’s hit weakly (.227 AVG and .318 SLG). The curveball is a great pitch for Bedard; no wonder he throws it so much.

I wanted to add a note at the end here about which pitches Bedard used to get his strikeouts. We have PITCHf/x data for 50 of his 221 strikeouts. Of those 50 K’s, 22 of them came on the fastball, 21 on the curveball, and 7 on the cutter. That lines up pretty well, percentage-wise, with his pitch mix with two strikes on the hitter.

Hopefully, we’ve learned a little about how Bedard dominated hitters in 2007–a strong fastball/cutter combo and an outstanding curveball. His changeup could use improvement, but it’s his fourth pitch, so that’s really a small complaint. It will be interesting to see if he can maintain the strong performance in 2008 as well as whether he will be doing so as part of the Orioles or on a different team.