College pitchers the better bet

Rob Neyer is on vacation. Two guest columns will run in his place on Thursday and Friday of this week.

In The New Bill James Historical Baseball Abstract (2001), James writes,

The most phenomenal fact of life in baseball today is that major league teams continue to use first-round draft picks for high school pitchers ... It has been obvious for twenty years that this is a stupid, stupid gamble ... yet every year, four to seven first-round picks are invested in these turkeys.

But if you think the issue is a dead horse, you're wrong.

The high schoolers have their advocates. In the spring of 2003, Jim Callis published a study in which he concluded that while collegians are slightly more likely to become major league regulars, high schoolers are about as likely as collegians to become stars, and significantly more likely to become better than average major leaguers.

High school pitchers have their advocates in front offices, too. James' estimate was a little on the low side; the last 10 drafts have averaged roughly seven high-school pitchers drafted in the first round. And with the amateur draft coming soon, I figured this might be a good time to look again at the issue.

The question is quantifiable, and when Bill James published his book about Win Shares, he contemplated using the method to study the draft. With the help of Tangotiger, whose work includes a thorough analysis of Win Shares, I devised a study that would bring some objectivity to the subject.

Win Shares is designed to measure a player's contribution to his team's performance. That contribution is translated to a "claim" on his team's victories. Each Win Share is worth a third of a win, so three Win Shares equals a win. Fifteen Win Shares (five wins) is a good season, and 30 Win Shares gets a player into the MVP mix (or should).

Our goal was to treat both classes of pitchers fairly. I plan to post (somewhere) a longer article, later this summer, that will explain our methodology in more detail, but here are its outlines:

  • We considered the first 10 high-school pitchers drafted in the 1982 through 1991 drafts, and the first 10 college pitchers drafted in the 1985 through 1994 drafts (the latter group starts three years later because college draftees are essentially three years older than high schoolers);

  • We looked at each pitcher's total contributions through age 30, because most pitchers will have peaked by then;

  • We allowed for "early bloom" years between the Age 18 and 21 seasons for high schoolers.

    Cutting to the chase ... This new study confirmed what Bill James found a long time ago: college pitchers yield a much higher return on the investment than high schoolers.

  • 65 percent of the collegians earned at least one Win Share through their Age 30 seasons; only 41 percent of the high schoolers did the same.

  • Seven of the top 10 Win Shares earners through Age 30 were collegians.

  • The collegians account for 64 percent of all Win Shares earned by these two groups (again, through their Age 30 seasons).

    The high-school draftees earned an average of 15 Win Shares through age 30. Their college counterparts earned 26 Win Shares, a 65 percent lead over the high schoolers. (That difference of 11 Win Shares equals roughly four victories.)

    Even if we consider only the pitchers who reached the majors, the collegians still do better: an average of 41 Win Shares for the college guys, 37 for the high schoolers.

    We broke the pitchers down by which slot they were taken. If a player was the first college pitcher taken in a draft, we put him in Slot No. 1; if he was the ninth, he got Slot No. 9. To make the data a little easier to handle, we broke the slots down into groups of three. The top three college pitchers went into the College 1 group, and so on.

    Group Win Shares
    College 1 1,302
    College 2 736
    College 3 420

    High Sc. 1 1,037
    High Sc. 2 415
    High Sc. 3 146

    College pitchers in Group 3 were just as valuable as the high-school pitchers in Group 2. What's more, the value of the first three high-school pitchers drafted (1,037) is right in the middle of college pitchers one through six (an average of 1,019).

    There were seven years of overlap in our study: 1985-1991. Based on what we found, of the first 15 pitchers selected in a draft, six of the first nine should have been college pitchers, and the next six should have been balanced between high schoolers and collegians. This would give us a 60/40 split. What we saw was that of the first nine pitchers taken in those seven drafts, 40 were from college and 23 were from high school. That's very close to what we would have expected (42/21). Of picks 10 through 15, 22 were from college and 20 from high school. Again, this is close to what we expected (21/21). Our conclusion is that from 1985-1991, teams' scouting philosophies and expectations regarding high school and college pitchers were in line with what actually transpired. We had equilibrium.

    What about individual performances? The top Win Shares earner by far was Greg Maddux, who was in the high school group. He finished with 205 by age 30. He's in a class by himself (and more about him in a minute).

    The next tier was split, two high schoolers and two collegians, and the spread among them was small:

    After that, the collegians dominated the field, with more of them earning 10, 20, 40, 60, 80, and 100 Win Shares.

    In almost every category, the collegians are the better bet. But does Maddux's performance show that the high-school pick might be worth the risk? He was, after all, easily the top Win Shares earner. And he and Doc Gooden were the only two pitchers from this collection of 200 pitchers to post a season with at least 30 Win Shares (they each did it once).

    In light of how much the collegians dominated this study, it would be hard to argue that a 0.5 percent chance of landing the one true Hall of Famer from this group would justify going with the high schoolers. Consider, too, that if we changed the parameters of the study just a little, moving the college draft class back a couple of years, that group would have picked up Roger Clemens, who easily would have been the top earner overall. He is also the last pitcher to net 30 Win Shares in a season. There's just nothing in the data to support the proposition that high schoolers yield a higher reward when they do make it, so they're worth the risk of a first-round pick. In truth, you're just as likely to land a Clemens as a Maddux.

    Based on this study, the collegians appear to give you two times the return on your investment, and the difference is most pronounced in the early picks. It's the same disparity Bill James found 20 years ago.

    Are high schoolers a stupid gamble? Occasionally a pick like Maddux, Gooden, or Glavine comes along, and for some organizations this might justify the risk. And since we found equilibrium in the overlap sample, it looks like the teams themselves, at least as a group, have figured this out. As for spending a first-round pick on a high schooler, it's worth noting that Maddux and Glavine were second rounders. It's probably true that some high-school pitchers are worth a first-round pick. But it's also probably true that nobody really knows which high-school pitchers those are.

    Keith Scherer is an analyst for Baseball Prospectus and an attorney with the Air Force JAG Corps. You can send him e-mail at kscherer@baseballprospectus.com