Lines don't predict anything. Lines are set to keep the betting basically 50/50 and assure that the house wins. Holyfield was a 25 to 1 dog against Tyson. That they sometimes mirror the outcome is nothing but a statistical probability. Lines do not taking anything into account except money. If there are 10x as many notre dame fans as purdue fans and they each bet on their team, the line will skew to a big notre dame blow out until notre dame fans quit betting on nd. has nothing to do with market confidence.
Just because you can find examples where the line was wrong, doesn't mean they aren't the best information available before a game. House may set the initial line, but if there's betting going on then it's the market that has the final say in what the line is. That takes all available information into account - injuries, changes in starting lineups, weather, etc etc etc.
The house only wins because their strategy is to take a small cut from the middle either way.
A few years ago, prediction markets like Intrade popped up and were semi-legal for a while. Intrade allowed people to bet on things like political elections. This market-based prediction was a more accurate predictor than any pundit or even phone polls. The reason is because anyone who thinks they have a better strategy for predicting results will bet money on it, which affects the line. People who are most confident in their information affect the line the most. Anyone who thinks the line isn't accurate will see an opportunity to make money, and collectively this activity provides the most accurate predictor of success that you can get anywhere.
If you don't believe lines right before a game are the best predictor, then make a bet (even on paper without sending your money in) on any game where you think the line is inaccurate. You may win - you have a 50% chance. But to prove the point, do it 100 or 1000 times to make it statistically significant. You will find that as you increase the number of games you bet on, your success rate at beating the line will approach 50%. All you have to do to get rich is prove that you know better than the line on 51% of all games over a large enough sample size. There are a few people out there who are really that good - but they earn their millions on razor-thin margins, and their activity makes it almost impossible for everyone else to consistently make money over a large number of bets (exactly the way a good market should function).
Take your example of the ND fans. If as you suspect, ND fans make up the majority of bets on ND games and always skew the line against NDs favor, then you are noticing a potential opportunity. You could bet against ND on every game for the next five seasons and if you parlayed part of each win into each of the next few games, you'd be retired by 2020. The problem with this thinking is the assumption that there aren't already people out there who capitalize on this. If you can think of a strategy like this, then so has somebody else.
I tested this theory by playing a friendly game of ESPN college pickem for about 7 seasons. The format is to pick the winners and losers each week, then rank them by order of confidence. I came up with a stupid-simple formula that mostly just ranked them by order of betting lines. I won my league every year, was in the top 1% nationally every year (with the exception of a couple where I forgot to make picks), and ranked as high as 13th nationally by the end of each season (with 200,000 entries). I can tell you I'm not that knowledgeable of a football fan, but put my trust in the market and it paid off every season because each season was a sample size of over 100 games. Each season, I would see this strategy put my ranking higher and higher nearly every week. If the season went on for 50 weeks instead of 14, then I would expect the only people who would be tied with me would be the other folks who figured out to do exactly the same thing.
In the end, I see this as all the more reason not to waste money betting, but to use the line as information when just curious who is most likely to win and whether it's predicted to be a close game. Let me know if you find a better predictor - I'd love to retire young too.
