The Differences Between Year Two and Year Three Models

Both models are utilizing market share statistics, which means a player is being compared to his teammates within those statistics. Across a large sample, there's not a meaningful difference when looking at raw numbers versus per game ones. The year two model is trying to predict what's going to happen in year two and year three of a player's career. But we have to remember that if there's an exception to the role, it doesn't mean the role is invalid.