Originally Posted by
unit1884
For example:
Chicago @ Vancouver
Chicago -1.5 pays +255
Vancouver -1.5 pays +240
Game goes into OT pays +295
All lines are from 5dimes.
If you bet $100 on each outcome, the results come to
If Chicago wins by 2 or more you profit $55
If Vancouver wins by 2 or more you profit $40
If it goes to OT you profit $95
If it ends up being a 1 goal game in regulation you lose $300
If someone has the odds of this happening (1 goal game in regulation) we can tell if this is profitable or not.
To equalize results under all winning outcomes one would bet:
$100 on Chicago -1.5 @ +255
$104.41 on Vancouver -1.5 @ +240
$89.87 on Game goes to OT @ +295
This would yield a loss of about $294.29 were either team to win in reg time by exactly 1 goal and a win of $60.71 otherwise.
This corresponds to payout odds of about -100* 294.29 60.71 ≈ .
So for this to be profitable, the probability of an exactly 1-goal victory in reg time by either team would need to be less than about 100 (485+100) ≈ 17.10%.
Historically, we have that of the 3,865 NHL games from the 2005/6 season through Nov. 4 of the 2008/9 season, 838 games (24.05%) were decided by exactly 1 goal (by either team) in Reg Time.
If we were to take this number as an accurate predictor of the 1-goal non-OT probability for this game (although given that it's just an across-the-board average we should certainly NOT expect this to be true in general across all ranges of relative team quality), then the expectation on this bet would be about (1-24.05%) * -1 ≈ -8.38%.
The point that bettors need to remember is that one can't simply take some group of -EV singles and expect a linear combination of them to be +EV. By betting the combination above and expecting it to be +EV what you'd be saying would be that you believed at least one of the bets taken in isolation to already be +EV (and Ito be clear, I'm not discouting that as a possibility).
As such, before considering a given bet combination one needs to consider which of the included bets one believes to be +EV, and then come up with a compelling argument as to why he believes that so, and then quantified. If such a bet can be identified and its edge quantified then one can use a staking strategy such as Kelly to optimize risk/return by potentially placing additional bets on the -EV outcomes as appropriate.
But just to be clear, unless one (or more) of the above bets is already +EV, there's no way that the combination could be anything but a long-term expected loser.