AxiomOfChoice
• Posts: 5761
Joined: Sep 12, 2012
September 30th, 2012 at 7:17:08 PM permalink
Hey Wiz,

I posted this to the sports betting thread a while ago: https://wizardofvegas.com/forum/gambling/sports/11254-correlation-between-total-and-point-spread-bets-in-football/

Some views, but no responses. I know, it's a hard question. I'm not even sure if there's enough data out there from real games to give a confident answer for anything but the most common spead / total combos.

Still, I think it's in interesting question. The answer would make a great appendix to your NFL betting section. Just sayin' :)
MangoJ
• Posts: 905
Joined: Mar 12, 2011
September 30th, 2012 at 11:29:49 PM permalink
I forgot the details of a model predicting sport result probabilities to answer your question, could you help out ?
AxiomOfChoice
• Posts: 5761
Joined: Sep 12, 2012
October 1st, 2012 at 12:22:40 AM permalink
Quote: MangoJ

I forgot the details of a model predicting sport result probabilities to answer your question, could you help out ?

If you had enough data, you could do it with past results, in the same way that the Wizard did it to evaluate the true value of a half-point off certain spreads, as well as the probability of a Wong teaser hitting.

Eg, I am asking, if the spread is 10pts and the total is 53, I want to know the true odds on an over + favorite parlay, as well as an under + underdog parlay. If you had enough past games where the spread was 10 and the total was 53, you could estimate the probability by looking at past results.

Now, admittedly, you probably don't have enough games where the spread is exactly 10 and the total is exactly 53. So, you probably want to organize your past data into "buckets" that don't cross over key numbers. Obviously if you make your buckets too big your model becomes less meaningful, and if you make them too small you lose data for each one (making your results less accurate). So, you might have to play with it a bit and see what works.

It may also be that you just don't have enough data to give something resembling an accurate answer, even for common spreads and totals.
MangoJ
• Posts: 905
Joined: Mar 12, 2011
October 1st, 2012 at 12:37:25 AM permalink
Ok, then all you must do is get a database and do the research. The by far larger problem is finding a book who will accept a correlated parlay bet.
AxiomOfChoice
• Posts: 5761
Joined: Sep 12, 2012
October 1st, 2012 at 1:11:10 AM permalink
Quote: MangoJ

Ok, then all you must do is get a database and do the research. The by far larger problem is finding a book who will accept a correlated parlay bet.

Not really. Some (most? all?) online books will accept them so long as the line is less than some percentage of the total. Obviously they understand there there is some correlation there, but they feel that it is not enough to overcome the house edge. First, I'd like to verify this. Then, even if, alone, it can't overcome the house edge, maybe if you combine it with a line that it a little slow to move (or a little off for some other reason) you may have something.

To be honest, though, the math interests me more than the small amount of action I put in at online sports books. Even if it's not possible to ever make a dime off of it, I'm still interested.
sodawater
• Posts: 3321
Joined: May 14, 2012
January 9th, 2014 at 12:00:30 AM permalink
to confirm, every PPH and offshore book I have seen will accept these bets.
KeyserSoze
• Posts: 413
Joined: Jul 14, 2013
January 9th, 2014 at 7:05:59 AM permalink
Many books will not accept a Favorite and Over, or Dog and Under parlay if the point spread exceeds 30% of the total.

Example:
Fav -17 and Over 48
Dog +17 and Under 48

Both examples would be considered correlated parlays and would be rejected.

Of course, every book is different. Many local bookies will take them. I went an entire season several years ago betting nothing but these types of parlays with a local book. It produced a modest profit.
Talent hits a target no one else can hit; genius hits a target no one else can see.
Wizard

• Posts: 26732
Joined: Oct 14, 2009
January 9th, 2014 at 8:33:36 AM permalink
I think this has been discussed here before.

I looked at this years ago and found there was not a strong correlation in games with a huge favorite. However, there were also so few such games that sample size error may have obfuscated such a correlation.

Personally, when I do see such situations, and some sharp bettors I know betting them, I sit on the sidelines with skepticism.
"For with much wisdom comes much sorrow." -- Ecclesiastes 1:18 (NIV)