Our main thesis was to discover whether favorites or underdogs were the better play. The results were clear: underdogs are generally the better play. Teams below .500 saw an increase in win probability of roughly 8.4%. Meanwhile, teams .500 and above only saw a 5.5% probability bump from the promotion. The strength of the relationship between a team’s winning percentage and the probability bump they received by applying the promotion is high. That’s reflected in a correlation coefficient of -0.35. In effect, that means the worse a team was, the better a pick they were for this promo.
Our main thesis was to discover whether favorites or underdogs were the better play. The results were clear: underdogs are generally the better play. Teams below .500 saw an increase in win probability of roughly 8.4%. Meanwhile, teams .500 and above only saw a 5.5% probability bump from the promotion. The strength of the relationship between a team’s winning percentage and the probability bump they received by applying the promotion is high. That’s reflected in a correlation coefficient of -0.35. In effect, that means the worse a team was, the better a pick they were for this promo.
Thanks Yisman. That is useful but it is just back testing. I'd kind of like to know if there is a math approach to predicting the value.
I can do it, I don't have the disposable leisure time right now to prove out the math.
I personally think it's advantageous for the short underdog player up to +6 points.
To prove that this is how I would go about it, I would use a sample size of ten years or greater, using SQL or SQL for Sports Analytics and isolate all NFL games with a closing line of six points or less.
From there I would identify from those games which underdog (you can get this information from the game summaries for instance) had a lead in the game at any point of seven points. It's a process, I don't have to explain further you get the gist of it.
Once you have all your data, crunch the numbers, come to your own conclusions.
If you used the Giants Monday at (pick) they would have paid you off as they had a seven point lead before losing outright.
Bengals last night at +145 or +150 was a loser, they never led by seven.
I would like to think it's not a bad play for short dogs you like, I will prove if it is or not when I get a chance using my 10 years as a sample size.
I do this type of shit for a living, life insurance actuarials, life expectancies, boring to 98 percent of all humans, not to a numbers geek however.
Good link
*edit*(for some reason link is not posting)
Google SQL for Sports Analytics From there, google playing numbers.
Thanks Steve, that makes sense but is still requiring collection of historical data.
I'm not sure if I am asking the impossible or not, but can this be predicted without needing to start with historical records? Just based on lines and odds. And be predictive.
And then back testing afterward to see if the prediction matches past results.
Dog, because you can win a whole lot more. Yesterday was +170/-200 and assume $50 limit. 50/80 or 50/25. No brainer (even though it went the other way yesterday), considering the whole idea that you might win either way.
Only reason to do favorite is if the max allowed bet is way above what you'd be willing to risk.