Examining the MLS Draft: Part 1 - Define Test and Success
Exactly what is a draft pick worth? Given that DC United just traded Brandon Prideaux for two third round picks, it seems that this is the kind of question that we would want to have answered, yet there seems little good analysis on the subject out there. Anyone can tell you that Taylor Twellman was a good pick for the New England Revolution, or that Andrew Mittendorf (Colorado's 1999 first round pick, #8 overall) probably didn't help Colorado too much. In an attempt to answer this question, the DCenters has examined the results from the first four rounds of every college/SuperDraft since 1999 in order to attempt to determine how successful, or unsuccessful, a draft pick tends to be. In this multi-part series, we will describe how we assembled and looked at the data (Part 1), how well the Draft works as a whole (Part 2), and compare DC United's success rates to other teams, especially in the Nowak era from 2004 onward (Part 3). The answer, in short, is that the MLS Draft ain't the place to reliably find talent, or even regular production. DC United has some success, but perhaps not in the most inherently obvious way.
In attempting to answer this question, it became clear that determining an objective test for a "successful" pick might be difficult. How much production should you expect from a forward, or a defender? How do you measure it? For our purposes, we sidestepped this issue, and attempted to incorporate the judgment of others: We looked only at MLS games played as a proxy for success. This offers us two benefits: This data is quantifiable, and to some degree it allows us to side-step whether or not someone is productive by letting a coach decide for us. After all, if someone isn't the best option, a coach would be foolish to let them keep playing (yes... we know the flaw there...). In short, The DCenters assumes that the more games played by a player, the more productive they are for a given team. This may seem to penalize players that transfer early in their careers to Europe (e.g. Bobby Convey) but Bobby Convey only helped DC United while he played for DC United. In short, the test seems to be a good one, and Bobby Convey would certainly be productive enough for the seasons he spent in DC.
Pre-1999 data was removed in order to attempt to filter out any distortion of the data that might result from the youth of the league as a whole. Data limited to four rounds since that's the current length of the Superdraft. Note that we would occasionally have to modify the name of a player in the Superdraft to get it to match the data from Climbing the Ladder. The difference between "Jon Conway" and "John Conway" was always changes to match our line-up data from CtL. I think we've caught most of these, but I'm not 100% certain. Note that the MLS All-time player register was also referenced to clarify confusing data points.
After assembling the list of the 354 draft picks, including the overall pick number, team making the pick, and the round number the pick took place in, these names were matched against the CtL MLS Line-up database to determine the number of games each player has started or appeared as a sub in in MLS games only. Cup and Continental games were not considered for this analysis, since the real question is how much production do you get from a player as for the MLS title, not the other events. The number of games started (GS) and sub appearances (Sub) were totaled for each player to determine games played (GP). Based on this information, the number of GP was used to classify a player in one or more categories:
- Player did not appear in any games.
- Player appeared in at least one game.
- Player appeared in at least 10 games.
- Player appeared in at least 25, 50, 75, or 100 games.
The increments of 25 were used in order to approximate a season's worth of production, given injury time, national team duty, or general rotation. An "Average number of games per season since draft date" was also calculated, but this should be viewed somewhat suspicously given transfer activity.
Hopefully we've explained all the boring stuff about our methodology. While it may not be perfect, I think it's a reasonable way to look at the data. Knowing this, I hope you'll be interested in Part 2, where we'll look at overall MLS data.