I've never bothered to look, but I'd be shocked if late-season lines weren't better predictors of game outcomes than early-season lines across all sports. Like FW said, there's just more information for the market to work with. So, if that's your definition of sharper lines, then I'm pretty sure they're sharper later in the season, but as a bettor, you're faced with the same lack of information as the bookmakers.
If a modeler is over-dependent on current season stats, or if he can't predict the contribution of new players to a team, his model is going to suffer in early season. If his model is strong in those areas, he should thrive early season. In professional sports, a well done purely math-model should perform very well early season. I'm not so sure about college. It's not my area at all, but I'd think that amount of player turnover and athletes with no career stats would make it awfully hard for an early season math-model.
This is kinda an "I know a guy who knows a guy" thing, so take it with a grain of salt. But, I know a guy who knows a guy who consistently just destroys early season NCAA hoops and fb, like, at a ridiculous clip, for many years without a model. He's not clueless when it comes to the math, but it's not a mathy approach.