By the way, the reason why the crew does not question your "late bet" on the passline with odds, is that the casino is more likely to win that "late bet" on the passline wih odds, which is why "put bets" are also welcome. But just try to place a "dont bet" after the point is established because no casino will allow it because the casino is more likely to lose a don't bet on the point after the point is established.
Its funny you say that b/c the guy playing next to me was watching how I was betting. After about 5 or 6 rolls, suddenly after the come out roll he throws his bet down on the don't pass. The dealer immediately told him he could not play that bet. The guy looked at me and then when back to hopping the 9.
It might have been mentioned above, but this only works with a decent amount of odds. I think you are turning your come bet into about a 11% house advantage by placing it after the come out roll. You need a lot of odds to bring that down below the house edge of the place bet on the 6/8. So in the case of 3/4/5 craps, I think you would be better off just doing the place bet and forgetting about the line.
33.33% on the 4 and 10
20.00% on the 5 and 9
9.09% on the 6 and 8
(personally, I calculate a weighted average house edge of 18.79%)
1.41% on the Pass Line
Multiply your $3 flat times the number of come outs multiplied by the difference in house edges, and you should have your theoretical answer.
However, since you only wanted to play the 6 and 8, I assume boymimbo's calculations are more relevant to you.
However, since you were doing this to avoid having to play any points other than 6 or 8, the $0.33 gain can be offset by the avoidance of the loss that would occur from losing the other points (4, 5, 9, 10) more often than winning them. Mathematically, if the point were 4 or 10, you could expect to lose 2/3 (66.67%) of the time. If the point were 5 or 9 you could expect to lose 60% of the time. If we assume the line bet to be $3 (the odds bet is irrelevant because it has no house edge), the math works out to show that you would have a net loss of $2 per occurance on the 4 and 10 points, and a net loss of $1.80 per occurance on the 5 and 9 points.
You could expect a come out roll to be a 4 or a 10 6/36 of the time and a 5 or a 9 8/36 of the time. So, on a come out roll of a 4 or a 10 the "sacrificed" $0.33 gain would be offset by a $0.33 loss (6/36 x $2), leaving you even. On a come out roll of a 5 or a 9 the "sacrificed" gain would be offset by a $0.40 loss (8/36 x $1.80) leaving you up $0.07 each time.
In summary, by not betting the passline initially, you would sacrifice $0.33 per every come out roll except those that would establish a point of 4, 5, 9, or 10. On those rolls you would be even on the 4 and 10 and up $0.07 on the 5 and 9. That nets out to be about $0.22 for every come out roll including the 4, 5, 9 and 10.
What I don't recommend is placing all the numbers for $3. So you have the option of taking $27 across, or $15 across. While the monetary risk is less, the house edge is higher, so on average over the course of time, your loss is pretty much the same either way, but you have a chance to win more doing $27 across. I'd rather do that.