In this thread** Mr. Floppy** asks a question about whether a football pool should require picks to be submitted by Wednesday of Thanksgiving week.
I lay out the pros and cons in Reply#5, pointing out that there is very little difference between Wednesday picks and Saturday picks.
But oh, here comes snark-meister DSYoungEsq, who by his own testimony is not a statistician, to complain that my answer lacks “statistical rigor”, and is therefore nothing more than “opinion and guesswork”. This is roughly equivalent to the morons who complain that evolution is “only a theory”.
Mr. Young, would you like to think for a moment about what “statistical rigor” in response to this question would entail? (Or is it easier just to snark?) The three Thanksgiving games and the 13 Sunday games have point spreads. The advantage of picking on Wednesday is that you have 16 games to choose from instead of 13.
To establish a mathematical value for that advantage, you’ll need to find a table that converts point spread to probability of victory (assuming that one exists; if not, you’ll have to create your own from years of football results and point spreads.) Then you’ll need to calculate the expected value of points based on optimal picks (based on larges point spreads) among all 16 games versus 13 games.
And all that work will only get you an answer for this year. The results will change from year to year based on the point spreads of the Thursday games relative to the Sunday games. A complete, rigorous answer would involve a matrix of possible Thursday point spreads.
And even then, it will have little relevance, because as bhal points out in the linked thread, people don’t bet that way in real life. You have to pick outliers to have a chance of winning a large pool. Quantifying the chance of hitting an outlier combination from 16 games rather than 13 is mathematically intractable. It exists, but good luck in putting a number on it.
All of the above conerns the advantage of Wednesday picking. Countervailing, we have the advantage of Saturday picking, which is that you have three more days for additional information to develop.
What is the quantitative value of this? Well, I expect you to do a study about how often and by how much NFL point spreads change from Wednesday to Saturday, and then to translate that into probability of victory and then into probability of winning the pool.
What’s that? You can’t do it because you’re not a statistician? But I should have done it before I had the effrontery to post in the thread? Fuck you.
Even then, your answer would ignore the intabgible benefit to the bettor of having three extra days to read about and cogitate over his or her picks. In an efficent betting market, this advantage is worthless. So I suppose I should also have done a study as to whether football betting is “efficient”, even though economists haven’t been able to settle the question for financial markets in 100 years.
So given all of these issues, what should I have done? Should I have told MrFloppy I’d get back to him in five years? Or should I have done what any sane person would do–post a qualitative response and tried to keep it to a reasonable length?
Shove your snark up your ass sideways.
