Tweet
Do not submit a prediction score in this thread. It just explains the contest.
Each week a score of 100 is possible. Here is the formula:
If you don't like the formula, suggest something different. I won't listen.
Format for scores:
---------------------
BYU|X
OPPONENT|Y
---------------------
First line of the post must be BYU and then the numerical score prediction. The second line must be the opponent name and the numerical score prediction. The data are collected by a little script I wrote so if you make things non-standard it makes it more likely that your prediction doesn't get counted. You must use the pipe delimiter (not a lower case "l", a "|") ... the script scans for the delimiter. The script doesn't care that you prefer a different delimiter. Don't add any other punctation. I am not going to check time stamps for predictions so your on your honor for not posting a prediction ex post. BYU or a reasonable variation (my view of reasonable, not yours) on the team name must be used or the script will not recognize (and hence record) your offering. You can use call the opponent anything you want (in case you like to mock the opponent via your choice of name).
FAQ
-----
1. If a diffuse prior is used in Bayesian estimation is there an easy way to tell whether the prior is still having a large impact on the posterior distribution of the estimate?
Yeah, often in a situation like that you will get a long tail to the posterior distribution. So, you might get the center mass at 1 but very little mass extends close to zero but the right tail has significant mass even out to 9 or 10.
2. Is the preceding relevant to anything on the cougarstadium?
Yes.
Each week a score of 100 is possible. Here is the formula:
Code:
Points = 34*winner + 33*f(|MOVD|) + 33*g(d(predicted,actual))
Winner = 1 if the winning team is correctly picked, 0 otherwise
MOVD = margin of victory diferrential: Predicted (MOV) - Actual(MOV)
d( , ) = Euclidean distance
f(x) = 1 - 0.75(x/15) if x <= 5,
= 1 - 1.00(x/15) if 4 < x <= 10
= 1 - 1.25(x/15) if 10 < x <= 12
= 0 if 15 < x
g(x) = 1 - 0.70(x/20) if x <= 5,
= 1 - 0.90(x/20) if 5 < x <= 10
= 1 - 1.15(x/20) if 10 < x <= 17
= 0 if 20 < x
Format for scores:
---------------------
BYU|X
OPPONENT|Y
---------------------
First line of the post must be BYU and then the numerical score prediction. The second line must be the opponent name and the numerical score prediction. The data are collected by a little script I wrote so if you make things non-standard it makes it more likely that your prediction doesn't get counted. You must use the pipe delimiter (not a lower case "l", a "|") ... the script scans for the delimiter. The script doesn't care that you prefer a different delimiter. Don't add any other punctation. I am not going to check time stamps for predictions so your on your honor for not posting a prediction ex post. BYU or a reasonable variation (my view of reasonable, not yours) on the team name must be used or the script will not recognize (and hence record) your offering. You can use call the opponent anything you want (in case you like to mock the opponent via your choice of name).
FAQ
-----
1. If a diffuse prior is used in Bayesian estimation is there an easy way to tell whether the prior is still having a large impact on the posterior distribution of the estimate?
Yeah, often in a situation like that you will get a long tail to the posterior distribution. So, you might get the center mass at 1 but very little mass extends close to zero but the right tail has significant mass even out to 9 or 10.
2. Is the preceding relevant to anything on the cougarstadium?
Yes.
Comment