February 24, 2006
Lies, Damned Lies
Any Such Thing?
"There's No Such Thing As A Pitching Prospect" has become such a catch phrase that even pitching prospects are using it. Of course, the pithy phrase can't be entirely true. If not for pitching prospects, where would major-league pitchers come from? Does a stork, his beak shaped like Leo Mazzone's nose, swoop down from the heavens to deliver fully-formed pitchers to the majors?
Everything in prospect analysis is relative. Pretty much everyone agrees that some discount needs to be applied to pitching prospects. Baseball America isn't treating Mike Pelfrey like he's Justin Upton, and rest assured that we wouldn't trade Matt Cain for Trevor Plouffe. But figuring out exactly what the discount rate should be is something that hasn't really been resolved. Traditional prospect analysis almost certainly isn't discounting enough, and I've come to believe that Baseball Prospectus isn't discounting enough, either. Although the amateur draft has seen a substantial correction--perhaps even an overcorrection--pitching prospects are still treated in trade talks like they're black chips at the Bellagio.
PECOTA needs to tackle this question head on, and the results are going to be a little shocking. Before we get into the detail, however, we should take a step back and think about those things that make developing pitchers different than developing hitters.
This is the best understood and least disputed plank in the TNSTAAPP platform. PECOTA addresses the injury question by means of an Attrition Rate, which tracks substantial drops in playing time among a player's comparables. Attrition Rate is an imperfect measure--it will reflect a drop in playing time for reasons unrelated to injuries, such as being moved to the bullpen or a utility role, or being traded to Japan. Still, it ought to provide us with some useful information.
Following are the PECOTA attrition rates in 2010, five years out from today, for the hitters and pitchers ranking in Baseball America's Top 50 list. We've excluded players like Pelfrey and Troy Tulowitzki who haven't played a significant amount of professional ball.
Player Attr. Delmon Young 13% Brandon Wood 27% Jeremy Hermida 24% Stephen Drew 48% Lastings Milledge 38% Prince Fielder 15% Howie Kendrick 17% Andy Marte 23% Ryan Zimmerman 20% Ian Stewart 43% Conor Jackson 24% Jarrod Saltalamacchia 33% Andy LaRoche 27% Carlos Quentin 36% Nick Markakis 39% Chris Young 31% Joel Guzman 29% Felix Pie 35% Daric Barton 28% Billy Butler 28% Hanley Ramirez 35% Carlos Gonzales 26% Russell Martin 42% Neil Walker 35% Erick Aybar 12% AVERAGE 29% Player Attr. Francisco Liriano 30% Chad Billingsley 44% Justin Verlander 35% Matt Cain 47% Jon Lester 37% Scott Olsen 36% Joel Zumaya 58% Jon Papelbon 67% Bobby Jenks 48% Homer Bailey 56% Philip Hughes 37% Anibal Sanchez 43% Anthony Reyes 42% Mark Rogers 65% Adam Loewen 58% Adam Miller 64% Dustin McGowan 41% AVERAGE 48%There's room for debate about the attrition rates attached to particular players. PECOTA does not know about a guy's injury history--it can only make inferences, and some of its inferences are going to be better than others. Regardless, the overall trend is plenty clear: PECOTA's attrition rates are 66% higher for pitchers than they are for position players.
We can cross-check PECOTA's estimates against a slightly more tangible data source. A scan of MLB.com's transaction wire reveals that there were 64 pitchers who spent some part of the 2005 season on the 60-day DL, versus 44 position players. The way that the Disabled List rules are set up, teams will usually place any player on the 40-man roster on the 60-day DL if he suffers a serious injury, whether he's presently in the majors or the minors. Also, 40-man rosters usually contain about an equal number of pitchers and hitters. Thus, these injuries are drawn from a base of about (20 players x 30 teams) = 600 pitchers and position players, respectively. Doing the long division:
Hitters: 44 / 600 = 7.3% chance of serious injury per season
These numbers are broadly consistent with the Attrition Rate figures. A pitcher is about 45% more likely than a position player to be placed on the 60-day DL during the course of a major-league season. Note also that these are one-year numbers. If we multiply each of those numbers by five--and I realize that this is sidestepping a lot of discussion about the interaction of injuries from one season to the next--we come up with a 37% rate for hitters and a 52% rate for pitchers, quite close to the rates that PECOTA identifies.
In any event, it seems safe to say that a pitching prospect is 50-70% more likely to be injured over the near-to-medium term than a comparably regarded hitting prospect. This is necessarily going to have a rather profound impact on their respective valuations.
Following are the maximum number of innings pitched in any professional season for the starting pitchers in the BA Top 50:
Cain 192 Liriano 192 Zumaya 151 Lester 148 McGowan 148 Billingsley 146 Papelbon 144 Loewen 142 Reyes 142 Olsen 136 Sanchez 136 Miller 134 Verlander 130 Bailey 104 Rogers 99 Hughes 87 AVERAGE 139 MEDIAN 142A healthy major-league starter might be asked to pitch 200-220 innings over the course of a full season. With rare exception, however, pitching prospects do not accumulate more than 140 or 150 innings in a typical season. Some of this is because the minor-league season is a month shorter, but most of it is because teams, quite rightly, want to protect their premium young arms from overwork. Either way, a young pitcher promoted to a major-league rotation might be asked throw 50% more innings than he ever has before, and that's a lot to ask. That jump has to occur sooner or later for a pitcher to mature into a #1-#3 starter, but some guys are not going to make it, or are going to see their performance atrophy along the way.
What's more, the pitcher not only has to throw more innings once he hits the majors, but he'll have to throw more difficult innings. Major-league lineups filled with smart hitters will require more pitches to be retired, and the pitcher may need to turn to his breaking stuff or tertiary pitches more often, which means more strain on his arm. I believe that this "wall" effect is substantially responsible for the "injury nexus" phenomenon that Will Carroll and I documented three years ago. Our study was based on an evaluation of pitchers who had pitched effectively at the major-league level. Although our conclusion was that pitchers before the age of 23 were more likely to be injured, what we may really have been observing is that pitchers are more likely to be injured shortly after pitching their first full major-league season, because that season will involve throwing many more pitches than they ever have before.
If you've looked at enough Five Year WARP charts, you may have observed that many pitching prospects actually have their valuations go down over the course of their next five seasons. Put another way, PECOTA wouldn't trade today's Francisco Liriano for tomorrow's. This is partly a consequence of pitchers' high Attrition Rates: if a pitcher was healthy last year, we can say with more confidence that he'll be healthy in 2006 than in 2010.
While hitting prospects face injury and attrition risk--moreso than the TNSTAAPP cliché would imply--they counteract it by continuing to develop their talents up until the age of 26 or 27. Pitchers do not have this luxury. In fact, most pitchers flatline from about the age of 21 onward.
I went into our DT database, which covers all major- and minor-league pitchers since 1996, and evaluated the change in equivalent strikeout rates (EqK9) from one season to the next for pitchers of different ages. The ground rules for the study were as follows:
Here is the percentage change in strikeout rate (EqK9) according to this methodology:
Age EqK9 Change 20 3.4% 21 2.0% 22 -0.2% 23 -1.2% 24 -0.6% 25 -0.8% 26 -0.8% 27 -2.2% 28 -3.1% 29 -2.5% 30 -2.8% 31 -3.9% 32 -2.8% 33 -5.5%As we see, strikeout rates begin a slow, steady decline as early as age 22. This is easier to envision in the form of a career path graphic. Suppose that a pitcher's first professional season comes at the age of 19, when he posts an EqK9 of 6.0. This number is the major-league average, which would be a very impressive performance. Here's how we'd expect his strikeout rate to progress from there forward:
Now, there isn't a lot of downward momentum on strikeout rate until age 27 or so. But strikeout rates are certainly not improving, at least when we look at pitchers en masse, whereas a hitter might improve his productivity by 20-30% from age 21 to his peak. Strikeout rates aren't everything, and young pitchers can expect to see improvement in their command. Indeed, the pattern that success stories like Rich Harden have followed is to cut out a walk or two per game while holding their strikeout rates steady. But a pitcher certainly faces an uphill climb to get better overall when his strikeout rate is on the decline, and the good majority of pitching prospects can't do much better than hold steady.
This is certainly the most conjectural point, and I'm sure that it reads counterintuitively for people who are used to hearing about the dearth of quality arms. But there are a couple of strong inferential reasons to think that it's correct.
The first is related to the point that we've made above. Hitters, as a group, develop quite a bit during their early twenties, whereas pitchers as a group do not. Although the average pitcher in the International League is certainly much better than the average pitcher in the Midwest League, this is more a result of selecting out the better performers (this is, after all, the primary function of the minor leagues) than inherent, age-related improvement in the quality of individual performances. Think of it this way: the median age of a pitcher in the International League is 26, whereas the median age of a Midwest League pitcher is 22. But while 26-year-old hitters are quite a bit better than 22-year-old hitters, 26-year-old pitchers are not really any better than 22-year-old pitchers.
The second reason stems from an informed read of minor-league statistics. Following are untranslated strikeout rates per nine innings at the various professional levels:
Rookie League 8.4 Appy 8.4 Pioneer 8.5 Short-Season A 8.0 Northwest 8.2 NY-Penn 7.9 Full-Season A 7.6 Midwest 7.7 Sally 7.5 Advanced A 7.5 Cali 7.5 Carolina 7.6 FSL 7.5 Double-A 7.3 Eastern 7.5 Southern 7.4 Texas 6.8 Triple-A 7.1 International 7.0 PCL 7.1 MLB 6.4 American 6.1 National 6.6The next time that you hear about some minor-league stud who struck out a batter an inning in his professional debut in the Pioneer League, keep in mind that this is barely better than league average down there. This is one of the reasons that I believe that pitchers like Jon Lester and Jon Papelbon are overrated, at least on a purely statistical basis. People "read" Lester's stat line like they would a major-league stat line: He struck out 163 batters in 148 innings--that's outstanding. What kind of crack is PECOTA smoking?. Well, strikeout rates are 23 percent higher in the Eastern League (the most strikeout-friendly of the Double-A levels) than they are in the American League--and that's before accounting for the fact that the pitchers doing the striking out are a lot better in the AL than the EL. Striking out a batter an inning is not the benchmark for dominance in the middle minors--it's the benchmark for being on the prospect radar screen.
But we've strayed from our main point. Strikeout rates are one of the cleanest indicators of who is dominating the batter-versus-pitcher matchup, particularly since they are not highly impacted by ballpark effects. The evidence suggests that pitchers are getting the better of it more often in the minors, particularly the lower minors, and less often in the majors. A lot of minor-league hitters have holes in their swings, and either can't hit a breaking ball or don't have the bat speed to catch up with a fastball. These players are weeded out before they get to the majors.
We can observe a similar pattern in home run rates; this is the number of home runs hit per game in 2005.
Rookie League .80 Appy .80 Pioneer .80 Short-Season A .59 Northwest .63 NY Penn .56 Full-Season A .73 Midwest .70 Sally .73 Advanced A .84 Cali 1.02 Carolina .84 FSL .70 Double-A .80 Eastern .84 Southern .69 Texas .88 Triple-A 1.02 International .98 PCL 1.05 MLB 1.05 American 1.09 National 1.01The pattern here isn't quite as steady, particularly at the Appalachian and Pioneer League levels, which are played almost uniformly at hitter-friendly, higher-altitude environments. Power becomes much more a part of the game as we move up the ladder. There are a finite number of baseball players who can drive a decent fastball 390 feet; a minor-league lineup might feature three or four real home-run threats, and a major-league lineup seven or eight. This, too, is a consequence of players getting progressively older at each level, since power is the skill that shows the most significant age-related development. It's also the reason that we've always paid a lot of attention to minor-league home run rates, and that we're paying more and more attention to groundball/flyball numbers. In any event, pitchers not only have a higher mountain to climb, but a steeper path.
Next week, we'll answer the question: if there is such a thing as a pitching prospect, would we even know who he is? Stay tuned.