I've been betting on a site that occasionally has a promotion where if you bet on an NBA game, and the team you bet on (no spread, just win/lose) ends up losing by 5 or less points, you get a refunded bet.

I looked up how close regular season NBA games usually end, and roughly 25% are won/loss by 5 or less points. That means it's a roughly 12.5% chance my team will lose by 5 or less if they're equally matched.

Thing is, there's lots of lopsided lines. And I was curious, is it better to bet on a heavy underdog, heavy favorite, or a 50/50 type game?

My gut says to bet either on a heavy favorite (low chance they'll get blown out, and if they do lose it's much more likely to be a small loss) or use a random number generator and pick a 50/50 game (which is what I've been doing right now, since if 12.5% of the time I get a refund, with the vig I'm paying it's a +EV bet).

I have NO math behind heavy favorite vs. 50/50 bet, it's just my gut. I could be TOTALLY wrong and would love your guys input.

i would imagine you should just take a 2.5 pt dog.

I'm assuming your spread bet has 50/50 chance of winning/losing.

If your bet is refunded even if you win, 3 pt dog should be the optimal play as the push frequencies losing between 1 and 5 should max out at that spread. Maybe close to 15%.

If your bet is NOT refunded if you win, then a pick em game should be the optimal play, because that offers the best push frequencies for losing between 1 and 5, that is around 12.5%.

For example, you would not take a 20 point favorite because the push frequencies for losing between 1 and 5 are very low. And if your pick wins by 1-19, you lose your bet and you get no refund.

This is solely based on EV considerations.

Thanks for the post, appreciate it.

Betting on the spread isn't allowed. So I pick a team, if they win (by any amount) I win the bet...if they lose by 1-5, I get my money refunded instead of it being a lost bet...if they lose by 6+, then I lose my bet.

There are odds though, so I can't just pick a heavy fav and get even money on the bet. So a huge fav might be -350 and the other side +300, but if my team loses by 1-5 I get refunded.

Your comment about betting on a 50/50 type of game makes a lot of sense (I can usually pick ANY NBA game that day)....that'd maximize the times I lose by 1-5 and it's a push instead of a loss. Maybe actually picking the dog is the best idea? (like a 2.5 point dog since a LOT of the times they'll lose by 1-5...aka lots of refunds...plus good odds when they do bink a win?)

The more I think on it, the more I think you're right.

sorry i meant to say favorite.

Assuming the money line bet is close to zero EV, you should take a 3 point dog on the ML to max push freqs for losing by 1 to 5 points. 3 point dogs are about +135 fair, which is about 43% win rate.

So, losing by 1-5 is essentially a push. Your EV turns out to be the same as the sum of the push freqs for the promo, as the win value cancels out the loss value:

0.426 * 1.35 - (1.0 - 0.426 - 0.15) = 0.15.

I don't think you can do better than that, unless you can get a +EV money line bet on a small dog, i.e. +2 or +4.