Friday 11 December 2009

Archive - Producing Results From the MLB Draft: A Case Study

Let me state first and for the record. I am not a statistician or a sabermatician. What follows is no doubt flawed in a number of ways. It is a "quick and dirty" snapshot albeit it was definitely not quick work.

Some caveats are obvious and I''ll try to mention them as appropriate but no doubt you may spot some I don't mention. Please be aware that I am aware of the limitations of this sort of thing.

Also, I will admit up front that this will come off as a Blue Jays focused article in that the parameters I used are defined by JP Ricciardi's term as GM of the Toronto club (2002-2009), but it is not primarily an apologia for Ricciardi's work in this regard. I brought to this study certain hypothesises, and those were neither completely borne out nor disproven. Be that as it may, it was necessary to adapt some sample size to be studied and given my allegiances it was the obvious choice.

The question I presented to myself was this: Which teams did the best at producing major league contributions out of the amateur draft? One could have broadened the question to include all players developed by a given system including international signings.

But it would have consumed even more of my time to have reviewed every roster for Latin players and then track down whether each was signed within the window of time I was looking into. I did, however, take it as a general principle that in a study which begins with the 2002 season, not a lot of Latin signees within that time frame will have reached the majors and made a notable impact given that such players usually take longer in the system.

In order to do this, I needed a measure of contribution that was consistent across all players. Ultimately I chose WAR (Wins Above Replacement) primarily due to the fact that career WAR totals for every player are readily available at FanGraphs. So, a simple, yet time-consuming, cross reference between the draft records of each team and the player pages on FanGraphs produces a database of WAR totals, as of the close of the 2009 season, for each team.

A few of big caveats jump out right from the start: the first is draft position. When you see two teams that produced generally similar results, you should referance that against the relative average draft positions of the two teams. For instance, in my final rankings the Orioles and the Phillies are only one point apart. Yet the O's have had vastly better drafting positions over the seven years under study (I discount 2009 as irrelevant since none of those players have yet reached the majors). Thus one can safely assume the Phillies have done better work.

A second caveat is time. A team will have naturally gotten more WAR from a player drafted in 2002 than another team will have gotten out of a player, even a notably better player, drafted in 2006. For example: The Blue Jays get credit for 10.2 WAR from David Bush, and only 0.6 from Travis Snider. But there's no question who's the better player. That's why I say this is a snapshot of a work that's continually in progress. You'd have to go back to drafts in which all the players are no longer active to avoid this problem and that becomes a question of historical curiosity instead of current relevance.

Finally I want to point out that my ranking system is flawed. If I were, in fact, a statistician I'd know how to weight the results more closely to the production received, but I don't. What I actually did was award point in reverse-order fashion. That will produce some anomalies which I am uncomfortable with, but I'm unsure how to remedy. If any of you math geniuses out there want to see my results in order to more properly rank them, feel free to ask.

What I tried to do here is strike a balance between quantity and quality. Some teams, for whatever reason, pushed a lot of drafted players to the majors but many of them had little impact. So there had to be a balance between just arriving, and being an actual quality player. To do this I arbitrarily chose the figure of 2.0 career WAR. This works against some younger players (Travis Snider, as mentioned, stands at 0.6 right now) but it also helps define the situation, as for the great majority of players those below 2.0 are non-factors.

To illustrate why this balancing act was necessary, I refer you to the St. Louis Cardinals. they ranked No. 6 in all of baseball in putting players in the majors from this study group, with 26. But only four of those players have more than two career WAR. that's tied for last in all of the major leagues.

The other balance I thought was important was between total WAR of all players, and average WAR. That has to do both with properly recognizing teams that produce many productive players, and at the same time controlling for the fact that otherwise unproductive teams manage to strike gold with one superstar who carries his team up the list.

For example, Khalil Greene (hardly a superstar of course) accounts for a full 73 percent of all the WAR produced by San Diego Padres drafting during 2002-2008 (and I've actually heard some Jays fans suggest Kevin Towers would have been a good hire for the new GM but I digress). As a general rule, the best player produced by top ranked teams here account from 1/4 to 1/3 of their (drafting) teams WAR production.

So the final list as arrived at by factoring in four component lists.

1. Total number of players drafted who reached the majors, irrespective of contribution (obviously many players had negative WAR totals). This one rewards the ability to draft a player who's at least good enough to earn promotion to the majors.

2. Total number of drafted players with have a 2.0 or higher career WAR total as of the end of 2009. This one controls for #1 by rewarding teams for producing actual contributors and not just September/injury call ups.

3. Net total accumulated WAR total of all players (total positive WAR value minus negative WAR values). this should be obvious.

4. Highest average WAR per player who reached the majors. This one controls somewhat for both the first item above (since every one of those cup-of-coffee call ups waters down the average) and also for the tendency of one or two great players to cover the failures of the rest.

Each list was awarded points in reverse order (this is particularly weak in the case of the second list where many teams tie for the same position but it's the best I could do) and then the points totaled to arrive at the following rankings. some other observations will follow. (relative draft position used to break ties)

  1. Oakland A's - 114 points
  2. Boston Red Sox - 107
  3. L.A. Dodgers - 97
  4. Arizona Diamondbacks - 97
  5. San Francisco Giants - 92
  6. Toronto Blue Jays - 90
  7. Milwaukee Brewers - 86
  8. Tampa Bay Rays - 86
  9. Colorado Rockies - 84
  10. Detroit Tigers - 83
  11. Kansas City Royals - 78
  12. Minnesota Twins - 76
  13. Atlanta Braves - 73
  14. Texas Rangers - 71
  15. New York Mets - 67
  16. Cleveland Indians - 60
  17. Washington Nationals - 55
  18. Florida Marlins - 54
  19. Pittsburgh Pirates - 51
  20. Baltimore Orioles - 51
  21. Chicago White Sox - 50
  22. Philadelphia Phillies - 49
  23. L.A. Angels - 47
  24. San Diego Padres - 40
  25. Chicago Cubs - 37
  26. New York Yankees - 36
  27. St. Louis Cardinals - 32
  28. Houston Astros - 32
  29. Cincinnati Reds - 25
  30. Seattle Mariners - 23

Some observations:

  • Even with the controls, one player can dramatically skew the rankings. For instance, Zach Greinke makes up 54.4 percent of the Royals WAR total. Take him out and re-figure the averages and you get (without adjusting other teams point total accordingly) a Royals total of 40 points instead of 78, which is much more reflective of the rest of their work. Or consider the Padres, without Greene their 40 points drops to 25.
  • Teams who had one player make up 40 percent or more of their WAR totals include (in order from highest percentage): over 60 percent: Padres, Nats; over 50 percent: Marlins, Royals, Phillies, Orioles, Astros; over 40 percent: Mets, Angels, Reds, Pirates, Cards, Braves, White Sox.
  • The most efficient team in terms of average WAR per player? The D'Backs with Chris Snyder leading the way at at 13.6 percent.
  • 29 players have accumulated individual totals over 10 WAR. The Red Sox are the only team to have drafted three of these men. 10 of these players were chosen in the first 10 picks of their draft.
  • 11 of those players are over 15 WAR.
  • Of the 60 players taken in the top 10 picks from 2002 to 2007, 1/3 have failed to reach the majors so far.

The A's average first draft position was No. 20 (over the 2002-2009 window), the Red Sox were at No. 30, Dodgers at 22, D'Backs at 16, the Giants averaged No. 11, and the Jays come in with an average pick of No. 15. The Brewers also averaged pick No. 11 (with rounding). By contrast, before 2009 the Rays average first pick was No. 3.

I'd humbly suggest that if your favorite team landed in the Top Seven, you should be pretty pleased with the results. Red Sox, A's, Dodgers, and D'backs fans should be giddy.

I obviously went into this with the thinking that the Jays were being under-rated in this regard and I think that conclusion bears out. But I was surprised when I first considered this subject how well the Giants in particular had done, and how poorly the Cards and Cubs have done. Hopefully even this ham-fisted ranking will provide a fresh perspective on this subject.

No comments: