August 14, 2012

PITCHf/x Profile

Dissecting the Decline of Josh Beckett

by Dan Brooks

In 2007, Josh Beckett finished second in the AL Cy Young voting. He led the league with 20 wins. He was a 4.8 PWARP player, good for third in MLB. He struck out 8.7 batters per nine innings and finished with 194 punchouts.

Fast forward to 2012. Beckett’s win-loss record is 5-9; more importantly, his walk rate is up, and his strikeout rate is down. He’s been worth 0.5 PWARP, good for 174th in baseball.

What happened to the dominant pitcher who anchored a World Series-winning staff?

This article aims to blend scouting information and statistical data in a look at what has changed for Beckett in the last five years. It’s more an exploration of a pitcher’s approach and evolution over time than it is an incisive look at a particular question.

Josh Beckett’s stuff: a primer
Josh Beckett’s repertoire consists mainly of five pitches: a four-seam fastball, a two-seam fastball, a cutter, a change, and a curve. Early in his career, he rarely if ever threw the cut fastball. Instead, he overpowered hitters with a well-located fastball, as well as a knee-buckling curve. But there has been a considerable change in grade for these five pitches across the five-year span we are examining, and as a consequence, Beckett’s approach has changed considerably during that time.

Because we lack PITCHf/x data from the beginning of Beckett’s career, we lose the ability to get an early statistical look at his stuff. But classic Beckett—on the Marlins, and perhaps even with the Sox in 2007—was, according to BP’s Kevin Goldstein, “a classic power pitcher with a 70 fastball and 70 curve.”

For statistical analysis, we are going to examine data from 2008-2012; we’ve chosen to start at 2008 because that season is when the majority of PITCHf/x systems were first fully operational (the cameras were only partially installed in 2007). However, we want to note that Beckett’s repertoire had already begun to change in 2008, and this sample may cover only part of the transition.

(Note: The pitch types we use when we talk about this PITCHf/x data have been sourced from the PITCH INFO LLC classifications available at Brooks Baseball.)

The plot below shows a year’s worth of PITCH INFO labeled pitch classifications from Beckett with horizontal movement on the X-Axis and speed on the Y-Axis. We see that Beckett’s three fastballs—shown in gray (two-seam), black (four-seam), and red (cutter) are delivered at relatively similar speeds; the cutter is slower than the other two, but the sinker and four-seam are often thrown with similar velocity. For many pitchers, the two-seam is slower than the four-seam, which gives it a bit of extra vertical drop due to gravity; Beckett’s is thrown at about the same speed.

You can see that Beckett’s sinker is different from his four-seam in that it gets more horizontal run; the pitch has a greater “tail” on it that darts inside to RHH and away from LHH. The cutter (red) has the opposite effect, although in fact it is a fairly flat pitch with little actual movement, relying more on its deceptive quality to generate swings and misses rather than any actual cutting action on the pitch.

Beckett also throws two off-speed offerings. The changeup (blue, here) is a prototypical “power change” thrown with very little velocity differential from the fastball. Unlike many changeups, the pitch does not have substantial horizontal run or screwball action; it truly is a change in speed and little else. According to Goldstein, even in Beckett’s best years, the changeup was always “just sort of there” and peaked at a 5, but was “more than enough if he was coming at you with a 70 curve.” The curveball (yellow) is a sharp breaking pitch thrown at a much lower velocity. Early in his career, the curve could be described as a “hammer”—a 70 curveball that would mystify hitters. He could throw the pitch for strikes or bury it out of the zone.

There is some question about whether or not Beckett has added a slower, more slider-like variant to his cut fastball or a split variant to his change this year, but with little data on these possible outlier pitches, we are choosing to ignore them in favor of “the big five.”

Grading Out Beckett
Rather than concluding with this information, we thought we’d give you the new scouting grades of Beckett’s stuff up front. These grades were compiled after discussions with Kevin Goldstein.

Goldstein says that Beckett’s 70 fastball has turned into a 55/60. His once-biting curve has lost a similar amount of stuff: Goldstein reports that he would be “leery” of putting a 60 rating on it and might call it a 50-plus. His cutter, he explains, is at best a 40. Given how often he throws it, the cutter could be something of an Achilles’ heel for Beckett going forward.

Goldstein is not terribly optimistic about Beckett’s performance going forward and grades him as a no. 4 or no. 5 starter going forward.

What has changed to warrant these downgrades in Beckett’s stuff? A loss of velocity and a change in the usage pattern of the cut fastball, along with what appears to a problem with missing location.

Yearly usage patterns
Armed now with knowledge about Beckett’s stuff, we can start to ask questions about how that stuff has changed. First, we’re going to look at how his usage of his various pitches has changed both across years, across innings, and across counts.

Plotted above is the rate at which Beckett has thrown each pitch as a function of year. We think there are several trends in this graph worth noting.

Let’s start by examining what should be Beckett’s bread and butter: the four-seam fastball. The four-seam fastball was his most frequently thrown pitch from 2008-2011, though it has suffered a precipitous decline in usage in 2012. Beckett’s two-seam fastball rate had declined steadily in every season since 2008 until this season, when he began to use it at the expensive of the four-seam. Now, the two-seam holds a five-percent edge, 25 percent to 20 percent.

Now, let’s examine the cutter. Beckett used to use this pitch much more rarely: fewer than five percent of his pitches were cutters in 2008. He now throws over 20 percent cutters, but he threw nearly 15 percent in 2010 and a shade under 20 percent in 2011. So, he’s been throwing the cutter with some regularity for three years, and he experimented with throwing it as early as 2008 and 2009.

Beckett’s usage of off-speed pitches has also undergone some development. The curveball, which in 2008 and 2009 was his dominant off-speed delivery (composing nearly a quarter of his pitches), has sagged in usage to the high teens, with the substantial decline in curveballs perhaps coming with the increased usage of the cut fastball. Finally, while Beckett’s changeup has always seemed like something of a work in progress, he gradually threw more of them in each season until this one, going from throwing only around seven percent changes in 2008 to nearly 15 percent in 2011.

Put simply, Beckett’s pitch usage seemed like it was on some sort of stable trend until this year, when the four-seam rate cratered, the two-seam rate skyrocketed, the cutter again increased in prevalence, and the changeup regressed.

Inning usage patterns in 2012
Because the goal of this article is to give you a sense of Beckett’s stuff at this snapshot in time, we’re now going to break down his 2012 data by inning to get an idea of when he’s throwing his pitches.

Focus first on the blue line, which represents Beckett’s four-seam fastball. While it is one of his most-thrown pitches in the first inning this year, it shows a steep decline in usage thereafter that perhaps explains why its overall 2012 usage rate has cratered. Compare the blue line with the red line (two-seam fastballs) and notice difference: beyond the second inning, hitters can almost sit on the two-seam fastball and ignore the four-seam offering, since they’re nearly twice as likely to see the two-seam variant. Though the data is noisy, the cutter shows a similarly interesting trend across the game. Beckett is unlikely to throw it in the first inning, and its usage rate shows a substantial increase during his starts.

There is some room for debate about what’s going on here. On the one hand, it’s an example of a classic pitching approach: establish the fastball early, and then go to your other pitches. On the other hand, it’s confusing: What’s the use of establishing the fastball, which is still a decent pitch for Beckett, if it’s simply going to be shelved for the rest of the game? It’s possible that this is a mental thing. Without a 70 fastball/curve, Beckett could be worried about using the four-seam and attempting to pitch to contact with his two-seam and cutter instead.

The curve shows its own trend: a nearly 10 percent jump in usage from the first inning to the second. The first inning is almost all fastballs, with few hooks, but the curve comes back after that.

One takeaway from this graph is that it’s difficult to say what Beckett’s pitch mix really is. On the one hand, you’d think that Beckett would be a prototypical power pitcher who relies on a mix of fastballs, a biting hook, and the occasional change or cutter. But the only inning in which you really see that kind of stable usage pattern is the second. In the first inning, when Beckett has been absolutely hammered this year (10.42 ERA), you see an almost obstinate reliance on the fastball, with very few curves. By the third inning (and throughout the rest of the game), the cutter has supplanted the four-seam fastball. Such an identity crisis from the first inning to the later innings is truly odd.

Handedness and count usage patterns in 2012
The table below shows pitch usage by batter handedness and count in 2012.

We include this table partially for completeness, and also to give the reader another window into when Beckett is throwing his various pitches. For example, whereas the change is thrown almost exclusively to left-handed hitters, the curve is thrown to hitters on both sides (but especially when ahead of right-handed hitters). Note how the cutter is thrown to both lefties and righties, but especially to righties (nearly 30 percent in all cases). Finally, note how rarely Beckett goes to the four-seam fastball when ahead of right-handed hitters.

A lack of velo
Beckett’s velocity has shown a substantial decline over the past few seasons.

If we look at Beckett’s fastball velocity over the course of his career, we see that he averaged over 95 miles per hour in 2008 and now clocks in at below 93. His much-maligned cutter actually was thrown harder in 2008 than his four- and two-seam fastballs are now! His cutter velocity has shown a similarly progressive decline, down to under 90 mph.

There is another interesting trend in Beckett’s velocity, this time with his curveball. This trend is not seen with other pitches, so we plot it only for a single pitch below.

This plot shows the difference in curveball velocity when there are two strikes on the batter as opposed to zero strikes or one strike. Beckett has shown a tendency over the past few seasons to throw a “put-away curve” that is harder, sharper, and tighter. In 2012, this “put-away curve” is nearly two mph faster than his normal curveball. Is this a case of Beckett having only so many good curves in the tank and saving them for when they matter? Is it a case of Beckett being somewhat “lazy” on his zero- or one-strike curveballs and then bearing down more substantially on his two-strike curves? Or is it Beckett purposefully “humping up” his two-strike pitches? Ultimately, the data can’t answer that question, but it is interesting that this pattern only exists for his curve and not for his other offerings.

Location, location, location
To examine where Beckett has been throwing his various offerings, we turned to our sortable Hitter Profiles. The profiles show plate locations from the catcher’s point of view for each of Beckett’s three fastballs in 2012. These profiles are calibrated so that the regions around the edges of the plate are to scale; they do not extend infinitely. This gives us a good idea of where he is locating these pitches around the plate.

Taking the profiles against right-handed hitters first, there is a general trend for Beckett to pitch away. The four-seam, two-seam, and cutter all have hot spots middle away. The two-seam’s location is interesting; rather than try to saw off bats inside, Beckett is trying to backdoor the pitch by tailing it over the outside part of the plate. But he isn’t quite getting his two-seam fastball down, and many of them are left over the middle of the plate. It is unclear if he is intentionally throwing the ball here or not—Could he possibly be missing this badly? More study or mapping over his target locations—and perhaps a talk with his catchers—would be needed to look at where he intends to throw his sinker, but it’s unclear why he would leave so many of them up and over the middle.

Turning our attention to the locations against lefties, we see some substantial differences. While the four-seam fastball often catches a lot of the dish, the two-seam fastball is now being thrown off the plate away. But the most puzzling location is that of the cutter. Against southpaws, Beckett is throwing a lot of pitches middle-middle and middle-up, almost certainly missed spots. In addition, the cutting action tends to bring the ball back onto the barrel of the left-handed hitter’s bat.

Swinging (and missing)
Now that we know something about the where, what, and when of Beckett’s various pitches, we can turn to the results. First, we will examine when batters swing (and miss) at Beckett’s offerings.

The plot below shows overall swing percentage at Beckett’s three fastballs.

Batters swing a shade under 50 percent of the time at Beckett’s four- and two-seam fastballs, but they swing at the cutter over 55 percent of the time. There is a clear difference in how batters react to these pitch types. Given his change in approach this year, is Beckett now having trouble generating the same swing-and-miss stuff that he once was?

The next plot shows swings and misses generated by Beckett with all three fastball variants:

We were surprised at this result, given that Beckett’s most effective fastball at missing bats is overwhelmingly his cutter, generating nearly a 25 percent whiffs-per-swing rate. The other two pitches have more or less held constant across the sample period, without showing the substantial decline that might be expected from the velocity decrease. But how does this compare to the major-league average? In 2012, Beckett’s four-seam fastball whiff percentage ranks 178th of 246 pitchers who have thrown at least 100 four-seam fastballs, but his cutter comes in at 12 out of 59 pitchers who have thrown at least 100 cut fastballs.

How do the curveball and change stack up? The next plot shows swings and misses on the two off-speed pitches in Beckett’s arsenal.

Despite an overall loss of effectiveness as rated by Goldstein and a substantial loss in velocity, Beckett’s change and curve have held up reasonably well. When it comes to swings and misses, Beckett’s curve currently ranks 21st out of 55 pitchers who have thrown at least 100 curves—right in the middle of the pack. It’s too bad we don’t have any data on the very early Beckett 70-plus curve for comparative purposes.

Pitch-type linear weights
Our last analysis measures the run value of each of Beckett’s pitches using the linear weighted outcome of each pitch. We are aware of the potential problems here—Beckett throws different pitches in different counts; the defense behind Beckett has changed over the last five years, which affects the outcome of these pitches; different pitches are thrown in different innings, etc. However, we also believe that this is one of the best ways we have to measure pitch value.

The way this analysis works is to take the linear weighted outcome of each pitch (throwing a strike in an 0-1 count is worth X runs, giving up a home run in a 3-2 count is worth Y runs, etc.), divide by the total number of pitches, and multiply by 100 to get the run value of throwing 100 of each pitch type. To eliminate year-by-year variation from slightly different run values for each event in each year, we’ve chosen to use run values from 2010 for the whole analysis. We’ve done this for each pitch in 2012, plotted below.

We’ll be the first to admit that this is where raw results start to diverge a bit from the scouting information. Josh Beckett, in 2012, has a nearly -2 run/100 pitches four-seam fastball. Let us contextualize that for you. Suppose you were to take every starting pitcher in baseball and compute this same number for their four-seamers. Josh Beckett would rank sixth out of 188 starters who have thrown at least 100 four-seam fastballs. And if you were to normalize all their scores into z-scores, multiply by -10, and add 50 (which creates a sort of ad-hoc 20-80 scale), Josh Beckett’s four-seam fastball is a 73.6.

Huh.

However, let’s see what happens when we do the same thing with his other pitches. The curveball ranks 42nd out of 128 qualifiers, good for a 56.8 rating. The cutter ranks 50th out of 71, good for a 44.7. The two-seamer ranks 83rd out of 159, good for a 49.2. And the change is 67th out of 125, good for a 50.1.

These grades are plotted above. Average across the board, except for the sparkling four-seamer... which has been thrown progressively less this year, in favor of the middling two-seamer and below-average cutter. Maybe our scouting information wasn’t so far off.
 

Pitch Tags compiled from the PITCH INFO LLC database. Both Harry Pavlidis and Kevin Goldstein contributed to this article.

Dan Brooks is an author of Baseball Prospectus. Follow @brooksbaseball
Click here to see Dan's other articles. You can contact Dan by clicking here