i suppose they could remain 50/50 right to the last game couldn't they? Or maybe we aren't talking about the same thing? I don't know what is under the covers in FG's system in terms of how often it updates projections but I assume it must at least look at who the remaining games are against so the model has be updating for that if nothing else. The BR model, being a running average run differential method, updates with every game played.
But when I said "more accurate over time" i didn't mean to refer to the convergence toward an end-point but which system tended to give better predictions over multiple seasons. say at some arbitrary date of interest within a season like the ASB or Sept 1 - or even opening day, though that is probably pretty worthless in any system!