Page 1 of 1

Vegas odds and sports model

PostPosted:Thu Mar 10, 2016 9:32 pm
by Don
If you go to ESPN's home page right now, their black-box advanced metric, BPI, claims that there's a 44% chance the Spurs will win the NBA title (Warriors are at 40%).

I searched for Vegas odds and the first page I found was http://www.vegasinsider.com/nba/odds/futures/ which listed Spurs at 16/5 (Warriors are 4/5).

Now, even though the house takes a cut, I'm pretty sure these odds are so wildly in conflict with each other that, suppose Vegas was indeed wrong, you'd be able to statistically come out ahead every single time by betting on ESPN's BPI. I mean, according to them, the expected value for betting on Spurs would be 16/5 * 0.44 = 1.408. If you hedge your bets on some other teams, there's no way you'll lose money if you believe ESPN's odds are correct.

While I certainly don't take ESPN (or any other site)'s statistics as gambling advice, this brings me to one of the things that always bothered me about these random advanced metrics. It's actually not impossible to say that some guy figured out a better predictor of who'll win basketball or anything else you can bet on in Vegas, since Vegas has a house edge and their goal is just to make the money wagered as close as possible. A lot of these statistics you can get at least the results for trivial or no effort (ESPN's been promoting their simulated odds for a while now). This isn't even like one of those things where you're supposed to expect to shell out a lot of money for some insider tips on stocks or whatever. There's no shortage of guys who wants to tell you their latest advanced metric has totally solved some sports and is willing to share all the data for free to prove how awesome they are. Since it is clearly not against the law to make a living by betting on sports (some people do it successfully, though from what I understand it's not reliable enough to try to go big), it seems to me one of the way these advanced metric can gain more acceptance would be like if the guy running this also made a few million dollars on the side. Okay, it's not often where these advanced metrics predict drastically expected odds compared to Vegas, though it happens often enough since Vegas is putting odds to maximize revenue from the average joe so even if those odds are wrong, they don't care if some super advanced model knows that and make a few thousand dollars off that particular game because they get more than that back from the average joe.

I remember ESPN used to have this stat-head challenge where they picked like the 8 biggest nerds or whatever to bet on the Playoffs based on what their advanced model predicts. Now I understand that each year the outcome is fairly unpredictable, but you'd think if these guys really have an awesome system, then over enough years they should have a huge edge compared to even Vegas odds. Maybe that's why all you get for those challenge is nerd points. You'd think if you're really confident in your advanced stats you'd just bet for real.