May 16, 2012

Prospectus Hit and Run

Beckett and Hyde

by Jay Jaffe

Printer-friendly

On Tuesday—his 32^nd birthday, coincidentally—Josh Beckett fired seven innings of four-hit shutout ball against the Mariners, taking advantage of one of the league's weak-sister offenses to rack up a season-high nine strikeouts. The outing pared Beckett's ERA by exactly a run, from 5.97 to 4.97, and more importantly, it allowed him to put an embarrassing sequence of events in the rear-view mirror. The Red Sox had scratched Beckett from his May 5 start due to a stiff latissimus dorsi muscle; the decision was made three days in advance because the Sox wanted to prevent a minor injury from getting worse. On the day of his next turn, a report surfaced that Beckett had played a round of golf the day after the announcement—hardly beyond the pale for a pitcher between starts, but questionable conduct for a player who was supposed to be recuperating.

That night, Beckett was battered for seven runs in 2 1/3 innings by the Indians and booed off the field when manager Bobby Valentine gave him the hook. He was defiant about his time on the links: "My off day's my off day." For a pitcher implicated in last year's fried-chicken-and-beer-stained collapse, it was more controversy and more evidence of tone deafness, if not the deeper character flaws that certain members of the mainstream media can divine from the press box and locker room.

Including Tuesday's outing, Beckett has actually delivered five quality starts out of seven; his only other dud was his season debut on April 7, when he was battered for five home runs and seven runs by the Tigers in 4 2/3 innings. His 7.6 strikeouts per nine and 2.8 walks per nine are both within hailing distance of his career marks coming into the season (8.5 and 2.7, respectively). The strikeout rate is weighed down by an eight-inning, one-K performance in his second start, but since then he has 31 strikeouts in 29 innings. Meanwhile, his 1.9 homers per nine is 94 percent higher than his previous mark, inflated by that early-season outing; it ranks seventh in the majors, with some other established pitchers, including a teammate, in the group as well:

Beckett's homer rate almost certainly owes something to bad luck; his HR/FB rate is 19.1 percent, well above the AL average of 13.6 percent (both figures based upon the MLB Advanced Media data in our sortable stat reports).

Beckett has been one of the game's best pitchers since emerging as a force in 2003, his age-23 season. Though he's just 18^th in innings since the beginning of that year, he ranks fifth in strikeouts; among the 55 pitchers with 1,200 innings in that span, his 8.4 K/9 ranks fifth behind Johan Santana (8.9), Jake Peavy (8.8), Cole Hamels (8.5), and A.J. Burnett (8.5). Furthermore, he's sixth in WARP, and 10^th in FRA+, the park- and league-adjusted index that is to Fair Run Average as ERA+ is to ERA:

Despite those rankings, it's difficult to make the case that Beckett ranks among the true aces of the game, in the Halladay/Sabathia/Felix Hernandez sense. He hasn't strung together back-to-back seasons with an ERA below 4.00 since 2003-2005 with the Marlins and has only cracked the top five in Cy Young voting once (2007, when he placed second). Durability has been an issue, in that he has reached the 200-inning plateau just three times since debuting in 2001 and has made 30 or more starts just four times. Indeed, Beckett has shown a very strange pattern over the course of his career, in that he's been much more effective in odd-numbered years than even-numbered ones, a Dr. Beckett and Mr. Hyde if you will:

The difference in innings may not be as significant as appears; another 138 this season would run his total to 179 2/3—right around his average of 177 from 2008-2011—and match his total in odd-numbered years. Yet all three of his key peripherals are worse in those even-numbered years, his home-run rate drastically so. Check out the effect on the opposition batting lines:

In even-numbered years, batters have hit for an average that's 18 points higher, an OBP that's 23 points higher, and a slugging percentage that's 54 points higher. It isn't simply a matter of luck on balls in play; his BABIPs in the even years have been just seven points higher. When we get to the advanced metrics (which don't include Tuesday's game), the difference is night and day:

Seventy percent of Beckett's value has been accrued in odd-numbered years. His four highest WARPs, and five out of six, fall in odd-numbered years; his lowest mark in any of those is 3.1 WARP. Meanwhile, the only odd-numbered year among his bottom six is 2001, in which he made just four September starts. Given the relatively minor difference in his split BABIPs and the extent to which FRA—and ultimately WARP—adjusts for the quality of defensive support received, it's unclear why this pattern should persist, but the best guess is simply injuries. As noted before, Beckett has rarely risen to the 200-inning level, mainly because he's rarely avoided the disabled list over the course of a full season:

Early in his career, Beckett was beset by blisters, which forced him to the DL at least seven times from 2002-2007. The majority of those bouts fell in even-numbered years, as have his back woes. If we include the non-DL stretches in our database in which he was limited by injury, such as the 10-day stretch last September when he left one start early and was scratched from another due to an ankle sprain, the counts jump to 148 injury days in odd-numbered years, and 283 for even-numbered ones. The caveat here is that we have no data for 2001, but including the minors, the 21-year-old started 29 times and tossed 164 innings, reasonable enough totals for a pitcher his age that we can assume we're not missing much.

Beckett's odd/even pattern is reminiscent of another pitcher who at his best was considered an ace, and who had even more success over the course of his career: Bret Saberhagen. Saberhagen reached the majors with the Royals a week before his 20^th birthday in 1984, and he won the first of two Cy Young awards as well as a world championship in 1985, going 20-6 with a 2.87 ERA. He slipped to 7-12 with a 4.15 mark the following season, rebounded to 18-10 with a 3.36 ERA in 1987, and so on. He won a second Cy Young in 1989, got hurt in 1990, and made at least 30 starts just one more time in the next 11 years, missing two complete seasons at the major-league level (1996 and 2000) along the way, to the point that 1998 was the only even-numbered year he saw major-league duty after 1994. But with Saberhagen, the differences between his performance in even and odd years were somewhat subtler:

Saberhagen's ERA in even-numbered years was higher, and his availability more sporadic, but his peripherals weren't necessarily worse. The divide between won-loss records, which carried a lot more weight in their day than Beckett's do in his, was more drastic; he was 99-52 in the odd years, 68-65 in the evens. Beckett has followed a similar pattern, going 76-38 in odd years, and 52-47 in evens.

For Saberhagen, the biggest issue, aside from the injuries, was his defensive support:

Most of the difference in those slash stats is accounted for by the 17-point difference in batting average on balls in play; the isolated power is exactly the same at .105, and the difference in walk rate is just 0.2 percentage points across split, 4.4 percent in the odd years, 4.6 percent in the even ones. The difference in advanced metrics isn't nearly so stark as it is for Beckett:

Saberhagen racked up 1,016 DL days during his career according to the data provided by Pete Palmer and Gary Gillette in the 2005 ESPN Baseball Encyclopedia. That still stands as the record, so far as anyone can tell. Those days are distributed fairly evenly, 540 in even-numbered years (including 1996 and 2000), and 476 were in odd-numbered years. In the end, about 59 percent of his value was accrued in odd-numbered years, compared to Beckett's 70 percent.

Saberhagen was less of a power pitcher than Beckett is, and Beckett himself is less of a power pitcher than he once was. Circa 2007, his best season, and the first for which BrooksBaseball.net has data, his average four-seam fastball speed was 96.2 mph, and he got swings and misses on the pitch 9.2 percent of the time, whereas now the heater averages 92.2 mph and gets whiffs just 4.7 percent of the time. In 2007, Beckett threw four- or two-seam fastballs 65 percent of the time, whereas now he's down to 50 percent. Back then, his favored off-speed pitch was his curveball, which he threw 26 percent of the time, and which got whiffs 12.5 percent of the time. Now he throws the curve 15 percent of the time and gets whiffs 10.8 percent of the time, instead favoring a cutter that he throws 23 percent of the time, and which gets swings-and-misses on 19.1 percent of them.

Like many a pitcher in his 30s, Beckett has had to evolve, though the extent to which he's become a pitch-to-contact type has been blown out of proportion by that early start; take that one away, and his strikeout percentage (K/PA) is 22.8 percent, matching last year's mark, and a whisker above his pre-2012 mark of 22.6 percent. Controversies aside, it will be interesting to see whether the new, less power-oriented Beckett is able to stay healthier than the old model, or whether he'll once again fall victim to the even-year curse that has defined his career.