[BobJoeJim]

[ QUOTE ]
I was thinking about this the other day. But what about teams that you "know" will have seasons like Duke and UCONN in NCAAB this year -team that are pretty much going to have less than 7 losses for the season. Couldn'y tou effectively Martingale (gasp) bet teams like those and come up a winner? That is, of course, that your bankroll is able to with stand 7 losses in a row, or whatever, but you would think that the likelihood of that would be slim.

I'm sure that you get my idea here. Let me know what you think.

[/ QUOTE ]
I get the idea, but the math just doesn't work. What do you actually mean when you say you "know" that one team will win? Anyone who has ever followed sports knows that upsets can and do happen, even the most spectacularly unexpected upsets. So when you say you "know" a team is going to win, you don't mean it's 100% certain. You mean it's 95% certain, or 90%. Or something. When the moneyline is -12,000 (I saw this on one of those 40 point spread college basketball games, like what you're talking about, earlier this year), then for you to break even betting the favorite they have to win MORE than 99% of the time. Are you really THAT sure? Because I guarantee you the linemaker who set that line thinks there's only, say, a 98.5% chance, and expects to make a profit off of that one time in 70 that the miracle happens.

The idea is simple, bet on teams that aren't going to lose. The flaw is that you don't win enough on each bet to negate the upsets that can, and WILL happen eventually, if you keep trying to take advantage of games that appear to be sure things.

EDIT: This concept reminds me of the blackjack "system" where you just double your bet every time you lose, and so each time you win just one hand, you make back all your losses, plus some. You start with enough of a bankroll to let you double your initial bet, let's say 8 times, because the odds of losing 8 50/50 bets in a row is less than half of a percent. That's basically zero, right? It's a sure thing!

The system works every time except the last time. In the end, though, it consists of repeatedly making negative EV bets, on the sole reasoning that you are "very unlikely" to lose. But more likely than not you will eventually lose, since that's the definition of a negative EV bet in the first place.