World Series time! Enjoy Premium-level access to most features through the end of the Series!
April 10, 2012
Have Teams Used Extra Draft Picks Efficiently?
Last June, the Tampa Bay Rays became only the third team in the draft era to have 10 or more picks before the start of the fourth round. Their 13 selections topped the previous record of 11 set by the Montreal Expos in 1990. The Rays’ 10 extra picks were their consolation prizes after outfielders Carl Crawford and Brad Hawpe and relievers Rafael Soriano, Grant Balfour, Joaquin Benoit, Randy Choate, and Chad Qualls departed via free agency.
Teams began receiving compensation draft picks shortly after the abolishment of the reserve clause in 1975. Free agent compensation was established as a way to maintain the balance of talent within the league. Generally speaking, when a player left one team to sign a free agent contract with another club, the player’s former club could receive the first- or second-round pick of his new club in the next draft, as well as additional supplementary—or “sandwich”—picks depending upon the quality of the player it had lost. The literary and cinematic success of Moneyball has contributed to rising interest in the draft, and recent iterations of baseball’s Collective Bargaining Agreement have added more compensation picks than ever, affording a considerable advantage to teams who are able to maximize their selections.
Between 1978 and 1987, 19.5 percent of teams had at least one extra pick in the first three rounds of the draft. By 2001, that figure had risen to 30 percent. Overall, more than a quarter of teams had extra picks in the first three rounds between 1978 and 2001.
* * *
More than half (53 percent) of players ranking among the top 500 in WARP since 2007 have come from the first 100 picks of the draft. In theory, the more high draft picks a team possesses, the greater its draft class should perform. Historically, however, that hasn’t been the case.
We’ll use WARP earned within 10 years of the draft to measure the value returned by the top three rounds of each team’s draft class, and we’ll refer to that value as “WARP10” going forward. Additionally, we’ll look at how the WARP10 earned compares to the average value returned from that team’s picks in all drafts between 1965 and 2001, which we’ll call “expected value.” The expected value is highest for picks at the top of the draft and decreases as picks get further from the top, though the decrease is not exactly linear.
* * *
From 1978 through 2001, there were 643 team-year draft classes, and 43.4 percent of those had exactly three picks in the first three rounds. Classes with one or two picks made up another 30.8 percent, and classes consisting of four or more picks represented the remaining 25.8 percent.
On average, a class consisting of only one pick returned 4.48 WARP10. Going from one to two picks increased the average WARP10 by 47.8 percent to 6.62, and another 37.1 percent gain was realized by increasing the pick count from two to three. Interestingly, the addition of one compensatory pick actually decreased the WARP10 returned by 11.7 percent to 8.00. At five and six picks, the average WARP10 spikes to nearly 12 but drops off significantly at picks seven and on.
We can learn more about teams’ use of additional picks if we compare the WARP10 returned to the expected value of those picks.
Predictably, the expected value increases as the number of picks increases, with each additional pick being worth an average of 1.82 WARP10. That contradicts actual results, however, where the value returned peaks at five and six picks and drops off afterward.
Measuring the difference between the expected and actual WARP10 is a good place to start thinking about efficient use of resources. Teams with three or fewer picks outperformed their expected WARP10 by more than a full win. Four-pick classes fell short of expectations by nearly 20 percent, but five-pick classes beat expectations by 8.3 percent. Curiously, adding more picks to the class resulted in even worse performance against expectations, though the sample size dwindles toward the upper extreme.
The Minnesota Twins outperformed the expected WARP10 of their picks by 32.8 percent, leading all clubs with a gross difference of 72.77, followed by Seattle (72.76), Boston (67.25), Colorado (48.33), and Cleveland (43.29). The Rockies’ 48.3 percent difference was the best in the game, mostly due to having 15 fewer classes in the study.
At the other end of the spectrum are the Los Angeles Dodgers at -73.96. The Dodgers averaged just 3.66 WARP10 per class, the lowest among non-expansion teams and 1.51 WARP10 worse than the next-worst team, the New York Yankees.
Most Efficient Teams, 1978-2001
Least Efficient Teams, 1978-2001
* * *
When I first saw this data, my thought was that teams were drafting more conservatively as they acquired more picks, perhaps due to their draft budgets not increasing proportionally as compensatory picks were awarded. However, that theory was put to rest when I noticed that the percentage of high school players drafted increased as the number of picks grew.
Given the increased percentage of high schoolers (the most volatile of draft picks) taken by teams with five or more picks, it appears my initial theory was backwards: teams are taking more risks with their extra picks. That would go a long way toward explaining the unexpected drop in value returned in classes with more than six early-round picks. If true, we should find significant variance in the WARP10 totals of the classes with the most picks.
It appears that theory holds up better than my initial suspicion, though not overwhelmingly so. The extremely limited sample of classes with seven or more picks likely precludes us from being able to draw any concrete conclusions from the data, but it does seem reasonable to hypothesize that the reason for the disappointing returns of the larger classes is that teams became less risk-averse as the number of picks at their disposal increased.