Tuesday, February 23, 2010
When ADP beliesPosted by Paul Singman at 5:30am
Knowing the ADP of players is an important tool for fantasy owners to help them maximize value during a draft. If you can, for example, strongly suspect that a player will likely still be available in the next round because of his ADP, then you can maximize value by drafting a different player in the current round while still getting the original player you waited on in the next.
Waiting too much is risky and reaching too often is wasteful. The trick, as you might presume, is to find the balance between waiting and reaching, which, although it is certainly important, will not be the focus of this article. Instead I will reveal the times when—or rather the players for which—ADP can be slightly misleading. And to help illustrate my point, I would like to point our collective attention to an anecdote describing the 1989 New York City mayoral election found in the book Freakonomics:
In New York City's 1989 mayoral race between David Dinkins (a black candidate) and Rudolph Giuliani (who is white), Dinkins won by only a few points. Although Dinkins became the city's first black mayor, his slender margin of victory came as a surprise, for the pre-election polls showed Dinkins winning by nearly 15 points.
The conclusion of the authors is that a decent amount of voters must have lied in the pre-election polls, saying they were going to vote for Dinkins so as to not seem racist, meanwhile actually voting for Giuliani when the vote counted.
The way I intend to parallel this episode to fantasy baseball is not through the issue of race—I would argue baseball fans are relatively color blind when evaluating players, considering baseball does not have the same circumstance with any particular race that, for instance, football does with non-quarterbacking white players or hockey with black ones. Instead I would like you to think of the pre-election poll as a mock draft and then the actual balloting as a real draft. Clearly in the political landscape voters could not be trusted to tell the truth and knowing this, how confident can we be that the results of mock drafts will best reflect what will actually happen in our real drafts?
I would answer that we should be very confident and that ADP values from mock drafts is by far the best estimator we have of when a player will be drafted in a league before most fantasy providers start holding drafts. And once real drafts begin, the results of those are even more applicable to your league. The biggest problem with sites like Mock Draft Central, which by the way is a tremendous resource, is that their ADP values come along with a tremendous bias based on the order the players are listed in their draft window. If every site used the same order then this bias would not exist but unfortunately it does.
So far I've only talked about players on a macro scale. The next question to ask is whether there is any specific subset of players that can be expected to have larger discrepancies than normal in their mock draft ADPs and actual draft ADPs. Inspired by reader Jimbo's comment on this article over at Fangraphs, I would answer yes. Jimbo disclosed in the comment:
Guys like Soto, and even Cantu, are the sort that tend to fall below preseason ADP. At least in my league/experience. Mocks are one thing, but on draft day they’re among the first players teams wait on “that one extra round” while value pitchers or upside OF are taken. Relievers and catchers go much later than average, and so on.
Couldn't agree more. Although it is hard to get the hard data necessary to prove it is true that closers and catchers are drafted more aggressively in mock drafts than real ones because of the aforementioned bias inherent in ADP values, my personal experiences lead me to believe it is true.
First off, Mock Draft Central forces your mock roster to conform to norms, meaning you have to draft exactly one or two catchers and also the exact number of required pitchers. It does not require for a specific number of your pitchers to be relievers, however, people seem to have conservative approaches to mock drafts and take a standard two or three closers by default—a standard they may forgo in their actual draft. It is also common for people to finish a real draft without a catcher, opting instead to hold one more of their deep sleepers on the bench.
It is this forced roster conformity during mock drafts and also a greater sense of desperation to extract maximum value during real drafts (leading to more waiting as opposed to reaching) that leads to catchers and closers getting drafted slightly later than their ADP numbers would indicate.
I am not sure whether this theory has a practical application beyond simply adding a plus five or plus 10 (or whatever you think it should be) to closers' and catchers' ADP numbers, but what compelled me to write a full article on the subject is more the innovative thought process that goes into finding out small inefficiencies like this one than practical application. That is not to say though, that combined with the xADP model introduced in this article, one day we might be able to generate numbers significantly more accurate than standard ADP data of when players will most likely be drafted, which would be something rather significant.
Paul has been managing fantasy baseball teams for many seasons and writing for THT Fantasy over the past three years. He is currently a student at UPenn welcomes readers' thoughts at his email here or in the comments below.