This three-part series is designed to take an objective look at the data that we get from average draft position (ADP) reports, as the introductory post explains. Fantasy baseball draft research is only as good as the data that exists and the current system for generating ADP data from sites like MockDraftCentral is broken and in need of a new direction.
The general idea behind my fear that ADP reports are irrelevant is that the results which we see seem to only be a slightly altered version of the rankings that the mock draft sites present to the participants. At this current time in late January, the mock draft data is limited aside from 132 qualifying drafts from within the last week at MockDraftCentral. However, 132 drafts should be a fine sample to test out how closely the rankings much up with the ADP results. Because if my fear is validated then there’s a real problem here.
Using a Pearson Correlation, we can test the relationship between the ADP results from MockDraftCentral for each player with their default ranking from the site. There are various schools of thought on this but generally a coefficient above 0.750 is considered a strong relationship between two groups of data (0.00 represents no relationship at all while 1.00 represents a perfect relationship). When we analyze the Top 50, Top 100 or Top 180 players from the rankings, the correlations between the site rankings and the resulting ADP show an extremely strong relationship:
Top 100: .976 r value (95.26% variance)
|Average draft position matches site ranking almost perfectly|
Basically, the correlation between the ranking for a player and their ADP indicates a nearly perfect relationship for the Top 100 players and is still extremely strong when considering the Top 180 players. In other words, there’s very little difference between where a player ranks and where they are drafted on average (for instance, Andrew McCutchen is the 50th ranked player on MDC and his average draft position is 48.95). What this tells us is that ADP data from the site is extremely flawed as it is almost entirely linked to the site rankings. This holds true for results from other mock draft sites as well unfortunately.
Yet nearly every fantasy baseball site, including my own, that determines whether a player is a good bargain or not will reference MockDraftCentral’s average draft position data as if it actually means something. I’m guilty of this because there really isn’t any viable alternative. The concept of knowing a player’s average draft position is wonderful as it’s great to know where a player is being drafted versus their expected performance. However, if we go ahead and look at a player’s ADP on three different mock draft sites, we get three different values that may or may not be even close to one another. If ADP truly represented what the public thought of a player, that would not be the case.
Simply put, the data that comes from these sites is poor and not reliable. It’s something that could be so vital if it was truly usable data but the current system of mock drafting is what creates these poor results. In the next part of this series, we’ll take a look into exactly why this current way of doing mock drafts is tailor-made to create bad results.