I also had a very intriguing conversation with my Dad and we happened upon a great idea.
Essentially my theory (which is not backed up with data yet) is that a team won't cover the point spread for ever. A team might under-perform a long time if everyone is quitting or they are disorganized (applies more to football). Now a team in college basketball might win 30 in a row (Memphis 2007-2008) but I can be damn sure they didn't cover the spread thirty games in a row.
Because the spread relates to expectations instead of skill, the spread will grow and grow (what kind of growth? interesting question) until it surpasses even the best performance of a good team having its luckiest night. The Patriots went 16-0 bit didn't go 16-0 against the spread.
So essentially, a Martingale style system betting against a team to cover the spread could very well make consistent money long term because of the impossibility of a long, long losing streak. A team might go 30-0 but they won't cover the spread nearly that many times. And the advantage of the Martingale system is that it doesn't require you to win more than you lose, it merely requires that you win before you run out of money.
Now even ten iterations of the Martingale system (starting on a ten dollar bet ) exceeds $1,000. That is a pretty steep climb. But I would love to see what the chances are of a team pulling off a ten game spread-beating run.
I tried the Martingale system in roulette when I was younger. Obviously, my results were mixed.
But the Martingale system applied in a human-expectation system will have a natural counter built in because its humans who make the spread and they will over-expect. Turning the behavior of bettors against themselves.
Likewise, with enough statistical data, one could wait for a three or four game covering streak to start betting against that team to cover as a means of decreasing instances bet and decreasing the preceding "tail" of a run.