Correlated Parlays.

[jedi]

New to this sports betting thing still but I asked a question a while back about whether parlays were -EV and the answer was yes EXCEPT for some correlated parlays where you're betting a team and the over/under on the same game.

I THINK (correct me if I'm wrong) the consensus is to bet an underdog and the under, or the favorite and the over. The point being is that if the favorite covers, the over will be likely to hit and vice versa, so if you're inclined on betting the game straight up, you might as well go for the parlay.

Last week I won a parlay picking an underdog and the under. While the underdog won out, the point total came within 2 points of killing me. So, is this type of betting a bad idea? Are there other methods of making these bets?

I'm thinking of combining these types of bets with the bar stool pundit bets next week.

I started doing some NFL research on these a couple weeks ago but they ended up going nowhere. The best opportunities are Fave/Over - Dog/Under with high spreads, but Pinnacle for one doesn't allow same game parlaying with spreads of 10 or more. I'm guessing that's industry standard. The caveat is even these opportunities aren't all that hot.

[youtalkfunny]

A better example of a correlated parlay is "McNabb Passing Yards Over" to "Owens Receiving Yards Over". The outcome of one has a much more direct impact on the outcome of the other.

No bookie in his right mind would accept that bet.

[Indiana]

If we assume that the mean returns are similar for correlated and uncorrelated parlays then the correlated parlays should have less variability because they are closer to approaching a straight wager (which obviously has less variability than a parlay). I don't think that the mean returns should be all that different here should they?

Indy

Could you explain this a little differently?

[Indiana]

I could write up some assumptions and calculate the variance mathematically but this isn't the probability forum so I will refrain. What I am saying is this:

Let Y1=Random variable of gain from the correlated parlay
Let Y2=""""""""""""""""""""""""""""""" uncorrelated parlay

Then Var(Y2)<Var(Y1). Basically these two variables should have a distribution with a similar center, but Y2's distribution will be less flat. What does this mean? It means that you will achieve the mean sooner with the correlated parlays. Now str8 wagers are better mean returns than parlays right? So if you are doing parlays you are a gambler most likely and you should do independent parlays to perhaps achieve some of that upward variability like a blackjack or pai gow player does.

Indy

[Indiana]

I meant Var(Y1)<Var(Y2). Got the inequality backwards since correlated parlays are less variable.

Indy

If I understand you correctly, your assumptions about correlated parlays are right. The problem is the majority of combos are still -EV.

[Indiana]

I think any kind of parlay will be a smaller long term return over a str8 wager. To avoid any calculations, could you give me ur mean return across all correlated and uncorrelated wagers and also include the standard deviations? This would be educative w/o doing any math.

Indy

All I said in the beginning was most Fave/Over -- Dog/Under combo parlays aren't any good with respect to clearing the 55.56% threshold. I'm not sure what you're asking from me here.