September 13, 2005
Doctoring The Numbers
The Draft, Part Seven
Welcome back to my periodic series of articles--"periodic" in the sense that Halley's Comet is visible periodically--on the draft. This time around, I want to break the data down into tiny chunks to see whether certain positions make for better draft picks than others.
Using the technique described in the last part of this draft series, here's a breakdown of draft pick value for college and high school players, separated into pitchers and regulars, from 1984 through 1999:
Class 1st Rd 2nd Rd 3rd Rd Overall COL H + 31.2% + 28.2% +110.0% + 37.6% JUCO H -100.0%* - 22.0% +162.8%* + 4.6% HS H - 18.4% - 33.6% - 58.3% - 27.0% JUCO P +104.8%* - 18.6% + 61.7% + 53.9% COL P - 1.0% + 34.3% + 12.5% + 7.4% HS P - 36.5% - 2.4% - 19.3% - 24.6%In this table, each value represents the degree to which that subset of draft picks exceeded or missed expectations. For instance, collegiate hitters selected in the first round produced 31.2% more value than draft picks taken in those same slots in the draft would be expected to provide.
Asterisks are listed where the sample size was extremely small, fewer than 10 players. As the data for junior-college hitters can show, you can get some wacky data in small sample sizes. The data set for first-round juco hitters is made up of exactly three players, none of whom reached the majors. From now on, we won't even bother to list data for groups of five players or less.
As this data is for the entire period from 1984 to 1999, the conclusions--that college players are better than high school players across-the-board--is not surprising. Perhaps the most interesting new conclusion is that junior college pitchers are significantly more valuable than juco hitters, although both returned good value.
Here's the same data, but stripped to its relevant core, from 1992 to 1999:
Class 1st Rd 2nd Rd 3rd Rd Overall COL H + 21.3% + 50.7% + 30.4% + 28.2% HS H - 22.3% - 6.8% - 49.7% - 20.9% JUCO H SSS SSS SSS - 34.9% JUCO P SSS SSS SSS + 4.1% COL P - 19.6% + 14.3% - 36.9% - 14.6% HS P - 15.4% - 15.0% - 13.1% - 14.9%(SSS = Small Sample Size, less than five players; from this point on, we won't break down juco players by round--there aren't enough data points.)
The first thing that stands out is that almost every group of draft picks is underperforming expectations. There's a simple reason for that--as pointed out in Part 6, draft picks as a whole from 1992 to 1999 were worth about 15% less than picks from 1984 to 1991. Over time, more and more major-league talent comes from outside the confines of the draft, and that in turn reduces the value of the draft as a method for talent procurement.
We could normalize the data to account for this, but since the point of this exercise is to compare one set of draft picks to another, it really doesn't matter. Normalizing the data isn't going to change the fact that collegiate hitters outclass every other set of draft picks, in every round.
From 1992 onward, the margin between collegiate and high school pitchers is essentially a rounding error; there is no evidence whatsoever that high school pitchers are any risker than collegiate pitchers, whether in the first round specifically or for the first three rounds as a whole.
Finally, it's interesting to see that high-school hitters actually turn out worse than either set of pitchers. In Part 5, using a different study method, high school hitters were worth about 10% more than high school pitchers overall; by our new method, they're worth about 5% less. Since the main difference between the two methods is that we've added discounting to the new method, my guess is that the reason for the discrepancy is that high school pitchers render more of their value early in their careers--before they get hurt?--than high-school hitters do.
Junior-college hitters turn out the worst of the six groups, but in a sample of just six players overall...ignore that. That's the last we'll talk about junior-college players; there simply aren't enough players taken out of junior college to break the data down any further.
Next, we'll look at how players drafted at different positions fare. For instance, here's the data for every high school catcher taken from 1984 to 1999:
Pos Years 1st Rd 2nd Rd 3rd Rd Overall HS C 84-91 - 78.2%* - 55.2% - 73.7% - 71.8% HS C 92-99 + 18.8%* - 43.1% +136.1%* + 17.2% HS C 84-99 - 42.9% - 51.3% - 18.9% - 41.5% Years Biggest Bargains Biggest Busts 84-91 Todd Hundley, Greg Myers Tyler Houston, Kurt Brown 92-99 Jason Kendall, A.J. Pierzynski Joe Lawrence, Ben DavisBiggest bargains and biggest busts are simply the players who most outperformed or underperformed the expected value of their draft slot.
In the 1980s and early 1990s, high school catchers developed a reputation for being the place where draft picks go to die, and with good reason--they returned less than 30 cents on the dollar. Or to put it another way, over that timeframe they were about half as valuable as high school pitchers, themselves no bargain.
Something changed in the 1990s. Not one catcher selected in the first three rounds from 1984 to 1991 developed into anything remotely resembling a star, but in 1992, the Pirates took Jason Kendall with the #23 overall pick. Paul Konerko has had his moments, and Justin Morneau is climbing the charts--both no doubt helped by a position change--but the fact remains that all three players will almost certainly end up with better careers than any catcher taken between 1984 and 1991.
Moving on to first base:
Pos Years 1st Rd 2nd Rd 3rd Rd Overall HS 1B 84-91 SSS - 98.6%* - 42.4% - 56.6% HS 1B 92-99 - 30.8%* -100.0%* - 36.4%* - 52.9% HS 1B 84-99 - 29.0%* - 99.3% - 39.4% - 54.5% Years Biggest Bargains Biggest Busts 84-91 Rico Brogna, Reggie Jefferson Drew Denson, Lee Stevens 92-99 Derrek Lee, Nick Johnson Matt Smith, J.J. DavisWow. High school catchers get all the bad publicity, but high school first basemen have actually been a worse value for the dollar over the entire 16-year draft study, and whereas teams seem to have learned how to avoid making mistakes behind the plate, they have shown no improvement here.
In total, 39 high school first basemen have been drafted in the first 100 picks from 1984 to 1999. Just one of them--Derrek Lee--achieved stardom. The results of the second round are particularly grisly--of the 16 first basemen drafted in the second round, the most valuable proved to be the immortal Tim Hyers. Let's put it this way: Chris Weinke was one of the data points.
Why the terrible performance? It's pure speculation, but if you're playing first base on your high school team, you're probably not the most athletic player in the world. No doubt almost all of these players were drafted for their bats, but athleticism has a lot to do with whether that bat develops, particularly since even the best high school hitters need a lot of refinement before they're ready for the major leagues. It's not a surprise that Lee, the best player in this group, is a remarkably good athlete for a first baseman, what with his Gold Glove defense and 15 steals a year. It's that athleticism that has helped him continue to improve as a player into his late 20s.
But most high school first basemen, by definition, have "old players' skills." It's a well-established point of baseball analysis that players with old players' skills--players who take'n'rake but lack speed or defensive aptitude--peak earlier and decline faster than other players. An 18-year-old with old players' skills is generally not a winning combination. (The picture is very different at the college level. Very, very different. More on that later.)
Continuing the theme of high-school hitters at less-athletic positions, let's take a look at the comic relief that is second base:
Pos Years 1st Rd 2nd Rd 3rd Rd Overall HS 2B 84-91 NSS - 58.2%* SSS - 70.6% HS 2B 92-99 NSS - 87.2% -100.0%* - 89.7% HS 2B 84-99 NSS - 79.5% - 99.4%* - 84.0% Years Biggest Bargains Biggest Busts 84-91 Uhh…Alex Arias Mike Hardge, Glen McNabb 92-99 Uhh…Brent Abernathy Victor Rodriguez, Cleatus DavidsonIf SSS stands for Small Sample Size, NSS stands for No Sample Size--no high-school second baseman was ever taken in the first round. Victor Rodriguez, the #39 pick in the 1994 draft by the Marlins, comes closest. The reason for this, and the reason why high-school second basemen make such awful draft picks, is pretty obvious. At least at the high-school level, second base is a position by exclusion. Shortstops, center fielders, catchers...these are all skill positions, and the guys who man those positions for their high school teams are there because they're the most skilled players available. Players put at first base or in the outfield corners tend to be bigger, slower guys who are drafted for their bats. Third basemen are guys with good defensive skills--the ones who are drafted high probably could have played shortstop in high school--but end up at third base, in all probability, because they're considered "too big" to man shortstop.
Second basemen? The skills required to play second base are the exact same skills required to play shortstop, so if you're playing second base, it's because there's someone else on your team that can play shortstop better than you. If you're not the best shortstop on your high-school team, then unless you played with Alex Rodriguez, that's a pretty good sign that you don't have the skills to make it to the major leagues.
That is exactly what the data says. Twenty-seven high school second basemen were drafted in the first three rounds from 1984 to 1999; Alex Arias and Brent Abernathy are the only two to have exceeded their expected value. As a group, they returned just 16 cents on the dollar, easily the worst value of any class and position.
Compare that with their partners across the diamond:
Pos Years 1st Rd 2nd Rd 3rd Rd Overall HS SS 84-91 + 21.1% - 88.5% - 94.6% - 0.3% HS SS 92-99 - 7.4% - 41.2% SSS - 16.3% HS SS 84-99 + 9.9% - 64.5% - 95.8% - 6.7% Years Biggest Bargains Biggest Busts 84-91 Travis Fryman, Gregg Jefferies Patrick Lennon, Austin Manahan 92-99 Alex Rodriguez, Derek Jeter Matt Brunson, Mark FarrisRelative to other high school position players, shortstops have been a pretty good value. Obviously, the inclusion of Rodriguez and Derek Jeter has a strong impact on the numbers, but that's just it: There are superstars to be had out there. It makes sense that as high school teams are wont to put their best overall athlete at shortstop, guys drafted from that position have more ways to make it--with their bats (Gregg Jefferies, Gary Sheffield), their gloves (Royce Clayton, Pokey Reese), or in the case of Rodriguez, both.
Two interesting things that come out of the data. One is the fact that in the 1980s, the best high school shortstops almost all found success at points much further south on the defensive spectrum. Travis Fryman made it as a shortstop, but found most of his success at third base. In retrospect, it's borderline astonishing that Jefferies played shortstop in the minors, and Sheffield played 94 games there for the Brewers in 1988-89. Chipper Jones, like Sheffield, also quickly slid from shortstop to third base, then made a much slower transition to the outfield. Jay Bell was the only draft success from this period to spend most of his career at shortstop.
The best high school shortstops of the '90s, on the other hand, tended to stay at the position--Rodriguez, Jeter, Jimmy Rollins and Reese. (The other high-school shortstop from this era who found success? Michael Barrett.) It may just be a sample-size fluke, but I thought it was interesting.
The other interesting piece of data is that almost all of the success garnered by high school shortstops comes from the first round. Second- and third-round draft picks were almost uniformly awful. Forty-two high school shortstops were selected from the #31 pick on in our study; Rollins was the only one who exceeded his expected value by even half a (discounted) win. Most of the success was clustered not just in the first round, but in the first ten picks. Seven high school shortstops have been taken in the first six picks of the draft. Five of them are Rodriguez, Jeter, Sheffield, Jones and Dmitri Young. Corey Myers was inexplicably taken with the #4 overall picks by the Diamondbacks in 1999, a pick that was mocked openly at the time. The only real bust was Mark Lewis, who the Indians selected #2 overall in 1988.
All of the sample sizes are small when we break down the draft by position, so I don't want to read too much into the data. But it appears that while it's a good idea to invest in a high school shortstop if he's a truly elite prospect, getting the third- or fourth-best high school shortstop available is usually a losing proposition.
Pos Years 1st Rd 2nd Rd 3rd Rd Overall HS 3B 84-91 SSS - 26.3%* -100.0%* - 67.3% HS 3B 92-99 - 20.8%* +266.0%* - 72.4% + 31.1% HS 3B 84-99 - 34.5% +114.3% - 86.6% - 1.7% Years Bargains Busts 84-91 Scott Cooper, Dave Hansen Greg David, Gordon Powell 92-99 Scott Rolen, Eric Chavez Josh Booty, Kevin WittIn the 1980s, high school third basemen were nearly as bad a gamble as high-school catchers--Scott Cooper and Dave Hansen were the only two high school third baseman (out of 15) who even reached the majors. As with catchers, third basemen taken in the 1990s have provided a much healthier return. That's almost entirely the work of Scott Rolen and Eric Chavez. Brad Fullmer is the only other draftee to return more than a discounted win above expectations. In a sample size of just 27 players, two potential Hall of Famers can have an enormous impact on the rate of return.
The difference between the two eras is probably just noise, but contemplating Rolen and Chavez--and David Wright, a 2001 draftee who fits the same profile--I wonder if the increased rate of return is a function of high school teams making third basemen out of guys who might have played shortstop in an earlier era. Certainly, judging from their defensive prowess in the majors, all three could have easily played shortstop in high school. When deciding whether to draft a third baseman out of high school, that question--"could he have played shortstop?"--seems like a good litmus test.
Pos Years 1st Rd 2nd Rd 3rd Rd Overall HS OF 84-91 - 15.3% - 68.7% - 40.3% - 31.2% HS OF 92-99 - 44.6% + 19.2% - 82.0% - 31.8% HS OF 84-99 - 26.9% - 31.6% - 55.7% - 31.4% Years Bargains Busts 84-91 Ken Griffey Jr, Manny Ramirez Mark Merchant, Jeff Jackson 92-99 Carlos Beltran, Johnny Damon McKay Christensen, Jaime JonesThere's precious little to say here. Outfielders rank around the middle of the pack among high school hitters, and there's little variation in the data among years or rounds. It would be nice if we could break down outfielders by where they played, but the data we have is very inconsistent about listing a specific outfield position for each player. I find it personally interesting that the two best draft picks from 1992 to 1999 were made by the Royals; finding toolsy outfielders is apparently one of the team's only strengths. Then again, they also made first-round picks out of Juan LeBron and Dee Brown...
Let's run a chart ranking high school hitters by their overall value from 1984 to 1999, and again just for the 1992-99 period:
1984 - 1999 1992 - 1999 Pos Overall Pos Overall HS 3B - 1.7% HS 3B + 31.1% HS SS - 6.7% HS C + 17.2% HS OF - 31.4% HS SS - 16.3% HS C - 41.5% HS OF - 31.8% HS 1B - 54.5% HS 1B - 52.9% HS 2B - 84.0% HS 2B - 89.7%In essence: left side of infield is good. Right side of infield is very, very bad. And high school catchers, relative to other high school hitters, are a much better value than conventional wisdom suggests.
We'll look at a breakdown of college hitters next time.