What are the draft tiers — bust, role, starter, elite?
Four outcome buckets the model assigns probabilities across, cut by percentile of career value within each position.
Rather than predict a single number for a career, the model spreads its belief across four outcomes: bust, role player, starter, and elite. Every prospect gets a probability for each, and they sum to one.
The tiers are percentile cuts of career value within position. Elite is roughly the top of the distribution for that job, starter the reliable regular, role the rotational contributor, bust the bottom. Because the cuts are within position, "elite guard" and "elite receiver" mean equivalent things relative to their peers.
The reason to show four probabilities instead of one prediction is that the shape carries information a single number destroys. Two prospects can have the same expected value while one is a safe starter and the other is a coin flip between stardom and nothing. A team drafting for a contending roster and a team rebuilding should read those two players very differently.
Read the ends and skim the middle. The model has a real, measured track record telling elite outcomes from busts. Distinguishing "solid starter" from "rotational contributor" is close to a coin flip from public data, which is why those middle bars are greyed on the site rather than presented with the same confidence.