Basketball Teasers - Massively -EV

[Post-Oak]

[ QUOTE ]
If books aren't getting any action on basketball teasers then it is too bad someone doesn't have the balls to come up with a better teaser for the player. Say 5.5 point two-team teasers at -110 in the NBA (assuming that is still positive EV for the book). I imagine an offshore book could have gotten some new business with those odds.

[/ QUOTE ]

I did some rudimentary research and it seems like BBall teasers are so massively -EV that someone really should look into offering more points. They can then tout that they have the best basketball teasers anywhere, bringing in all kinds of suckers who don't realize how bad these bets are.

I looked at data from Massey's computer ranking site. The rankings he shows to the public now are useless as far as betting goes (because someone apparantely pays him for exclusive access to his margin of victory determined power ratings/predictions). He also no longer shows a calculated home court advantage, so the data is a little old, but it should still be instructive.

NBA (year/# of games/calculated worth of home court advantage in pts/winning pct of home team/average total score)

2005 1314 3.17 pts 60.244% 194.37
2004 1273 3.89 pts 62.156% 186.16
2003 1277 3.77 pts 62.866% 190.29
2002 1260 3.17 pts 63.840% 190.82
2001 1260 2.91 pts 59.683% 189.49

WNBA
2004 240 3.28 pts 61.667% 133.37
2003 256 4.23 pts 62.891% 136.31
2002 273 2.96 pts 60.806% 135.68
2001 274 2.82 pts 60.079% 131.11
2000 272 3.81 pts 60.294% 137.70

NCAA (~330 teams were tracked for each of these years)
2005 4778 4.07 pts 67.252% 137.40
2004 4674 4.09 pts 66.285% 138.01
2003 4722 4.20 pts 66.950% 139.56
2002 4850 4.46 pts 66.682% 141.78
2001 4724 4.56 pts 67.057% 142.07
2000 4707 4.20 pts 66.611% 140.34

Let's take the case where you are teasing teams down from 4 pt favs to pick 'em. In NCAA games, it looks like that 4 pts should give you abut a 67% winning pct (we are assuming the line is efficient). The chance of winning both games would be about 45%. At even money payout, you can see how -EV this is. Two things strike me as odd. Number one, most books don't even pay even money, but payout at -110! Secondly, you are getting an even worse deal if you tease an NBA or WBNA game.

Obviously, I have ignored the key numbers theory completely. Key numbers are not nearly as important in BBall as in football, so I don't think it would change much anyway.

Also, it could still possibly be +EV to tease NCAA games where the total is very low (say less than 110), but I doubt it.

[Post-Oak]

[ QUOTE ]
Also, it could still possibly be +EV to tease NCAA games where the total is very low (say less than 110), but I doubt it.

[/ QUOTE ]

It occurs to me that one way to check this (if teasing might be +EV here) is to just look at the moneyline for the game. As an aside, comparing equivalent moneylines for different point spreads would give an idea if any teases could be +EV.

If a game looked like this:

Princeton
at Columbia -4.5 -280

and you assume the moneyline is efficient, then teasing two such games would be +EV (and would be far better than a moneyline parlay).

However, I don't think you could ever see such a steep moneyline for a 4.5 point spread.

[Performify]

This was really good discussion and I thought it belonged in its own thread outside of YTF's well.

[TomG]

Above is the margin of victory distribution for the last few years of NBA games. As you can see it's pretty uniform. For a type of teaser to be +EV, we need to pick up enough value from crossing significant point spreads to compensate for the fact we need to win both wagers (e.g., crossing both the 3 and 7 in NFL). There may be some margin of victory distribution sub-sets in there (for example low scoring, low point spread games) that have higher peaks in them. Here's the Excel files that contain all that info if anyone wants to do some sleuthing...

NBA 2002-2003
NBA 2003-2004
NBA 2004-2005
NBA 2005-2006

[Thremp]

TomG,

I won't be doing any analysis till I return home, but that is the exact information I've been looking for.

Many thanks.

[Thremp]

Post-Oak,

The MOV of 1 is a "dead" number in bball. Also I'd imagine large faves/dog games cloud the numbers a small amount. If you can find NCAA data I think it'd be interesting to gauge the wider dispersion vs. the lower totals to see which had a stronger effect.