January 29, 2003
Last week, I provided a list of position player breakout candidates as identified by a Peter Gammons survey of major league scouts and executives, and evaluated them by means of our new projection system, PECOTA, which uses comparable player data to provide an objective estimate of the probability of a marked improvement in a player's performance.
In this article, I'll do the same thing for the pitchers that made the Gammons list. I've put the pitchers second for a good reason: It is much more difficult to isolate components of a pitcher's statistical record that imply that a breakout is likely. Certainly, pitchers can and do improve dramatically at almost any age throughout their 20s (Kevin Brown and Randy Johnson are two prominent examples), but they don't follow the same dependable age-related patterns of growth that hitters do, and the precursors that predict improvement are much more difficult to detect.
Rank on Gammons List, Player, PECOTA Breakout Score
#1. Carlos Zambrano, RHP, Cubs 18%
Carlos Zambrano has electric stuff, but it's premature to place him with Mark Prior and Kerry Wood as a trio of fire-spewing Cubs pitchers whose appearance, like that of Hades' three-headed watchdog Kerberos, would surely portend a coming apocalypse. While Zambrano's Breakout score is high, his Collapse score (a measure of the probability of a dramatic decline in performance) is much higher, at 34%; a 22-year-old pitcher with a high strikeout rate and a high walk rate is an extremely volatile commodity. Zambrano displayed improved control in a brief stint in the Venezuelan Winter League, walking just one batter in 15.1 IP before the political turmoil in that country cut the season short. His performance there may be part of what caught the scouts' collective eye.
#5. Jake Peavy, RHP, Padres 8%
Peavy, also aged 22, is likely to improve upon his 4.52 ERA of a year ago. But most of that will be accounted for by better luck on balls in play (he allowed hits on 32.4% of balls in play last season, an unusually high figure), and an ERA that falls better into line with his peripheral statistics. Peavy's Breakout score, which measures changes in his PERA, and doesn't give him credit for improvement in his luck, is quite low (8%).
The poor Breakout score is not so much an indictment of Peavy in particular, who is a fine young pitcher, but of the greater risk associated with any pitcher at his age. Any of us would rightly prefer a 22-year-old hitting prospect over a 25-year-old hitting prospect. The exact opposite is true of pitching prospects; the older pitcher is the better risk. The reason has something to do with what Ron Shandler and other analysts have called the "injury nexus." The ages of 21 through 23 are a period during which a great number of pitchers suffer a catastrophic injury. Those who survive have been selected out on the basis of their durability, whether owing to superior mechanics, intelligent usage patterns, genetic fortitude, or luck. The risk of injury is such a preeminent concern when evaluating young pitchers that the mere feat of surviving the injury nexus intact is one of the most powerful recommendations that a young pitcher can have.
I tested the injury nexus theory by prematurely aging Peavy by exactly three years and re-running his PECOTA projection; all of his other statistics and attributes were left intact (as a college sophomore, I tried a similar trick with a scanner and an early build of Photoshop, but without so much success). The 25-year-old Peavy had a Breakout rate that jumped from 8% to 25%, and a Collapse rate that fell from 22% to 14%.
#11. Casey Fossum, LHP, Red Sox 17% #15. Tony Armas, RHP, Expos 18% #19. Danys Baez, RHP, Indians 12%
Fossum had the highest strikeout rate of the group, but wasn't quite as good as his 3.46 ERA from one year ago suggests. If there's a particular concern, it's that he spent the first half of the year in the bullpen. That isn't a bad thing in and of itself, and managers like Earl Weaver made a living out of developing young pitchers in just such a fashion. But PECOTA attributes slightly more risk to a pitcher who has yet to withstand a full season in a major league rotation. While his Breakout rate is solid, Fossum's Collapse rate (18%) is considerably higher than that of Armas or Baez.
Armas, by contrast, has already thrown more than 450 major-league innings at a level slightly better than league average. Although his strikeout rate isn't quite high enough to suggest that he's a superstar in the making, his comparables list includes a surprising number of high-profile names, such as Jim Palmer, Jack McDowell, and Mark Gubicza--all of whom turned the proverbial corner after a few seasons in the big leagues, and leveraged their experience into sustained success. (Based on his repertoire and his physical appearance--Armas is a big, strong guy--Black Jack looks like the best comp). It can't be said enough that any pitcher is risky, but Armas has a low Collapse rate (12%), and is a better risk than most.
PECOTA registers an ambiguous opinion with respect to Danys Baez. For one thing, it assumes he'll be a starter, while the Indians appear to be intent on using him as a closer. Baez has allowed an unusually small number of home runs for a pitcher who is not particularly adept at keeping the ball on the ground (his career GB/FB ratio is exactly 1.00). That and his intermittent command problems could undermine his ability to capitalize on his fastball. Eric Gagne's successful conversion to the closer role last season is sure to spark a blitz of cheap imitations, but Baez is a different type of pitcher, and a long shot to duplicate his success.
#8. Brandon Duckworth, RHP, Phils 14%
Brandon Duckworth, at 27, is the elder statesman on Gammons' list of breakout pitchers. It's almost certain that his numbers will improve next year; pitchers who strike out more than a batter an inning don't normally post ERAs over 5.00, and while Duckworth can be wild at times, it's not like he's Brad Pennington or Nuke LaLoosh.
Gammons' comment that Duckworth ought to throw more first pitch strikes is prescient, but not so much because Duckworth needs to lower his walk rate. Rather, Duckworth allowed an isolated slugging against of .316 when he was behind in the count last year, a rate that effectively turned every opposing hitter into Brian Giles. Duckworth's comparables list isn't exactly stuffed with success stories, but does include names ranging from Ken Forsch to Diego Segui, pitchers whose success was largely dictated by their ability to keep the ball in the park.
#12. Kyle Lohse, RHP, Twins 20%
Lohse gets the highest Breakout score of any pitcher on the list, but he also has the most ground to make up. Lohse has a home runs allowed rate that is nearly as poor as Duckworth's, but doesn't have the stellar strikeout numbers to make up for it.
In other words, Lohse could improve substantially, and still not be a heck of a lot better than league average; the most successful pitchers on his comparables list are Rick Rhoden and Todd Stottlemyre, pitchers who were not known for their dominance so much as their dependable mediocrity. Certainly, there are things to like about Lohse: his strikeout rate has inched upward throughout his career--especially in the second half of last season--and he's pitching in front of the sort of defense that should buoy his confidence. But it's tough to imagine that he will emerge the winner of a head-to-head battle in spring training with strikeout fiend Johan Santana.
The breakout scores generated by PECOTA are useful tools for evaluating a player's value going forward. In certain instances, such as those of Peavy or Jeremy Giambi, the weight of evidence provided by the comparable players data is sufficiently strong that it can lead to a more enlightened view of player development.
The PECOTA program, however, can only make assessments on the basis of the information that is provided to it--a player's record of performance in various categories, his usage history, and easily-measured physical attributes such as height, weight, and handedness. It cannot provide context where none is present, and it cannot evaluate latent traits that a scout might observe, but have yet to translate into improved performance.
The Breakout scores are driven by a half-century of empirical data, and over the long run, ought to be roughly accurate; one out of five players with a Breakout score of 20% will go on to have a breakout season, and so on. PECOTA has no way to distinguish the one player from the other four; a good scout might. Scouting and statistical analysis are sometimes presented as opposing approaches to player evaluation, but there is no reason why that need be the case. Baseball has reluctantly entered the information era, and the ability of teams to use both types of analysis in concert with one another will be a key driver of success.