tag:blogger.com,1999:blog-38600807.post2553897559033775338..comments2023-11-05T04:16:44.937-05:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Week 11 Game ProbabilitiesUnknownnoreply@blogger.comBlogger13125tag:blogger.com,1999:blog-38600807.post-19987246012645323972008-11-19T15:07:00.000-05:002008-11-19T15:07:00.000-05:00OK. I see what you are saying. Let me try it anoth...OK. I see what you are saying. Let me try it another way. Using your example of 83% win probabilitiy, this equates to a fair money line of about -488, or using the "dog", +488. If I look into my NFL database, and grind the numbers (assuming an average expected total), this equates to a spread of about 10.5 (not sure how you arrived at 7.73, but this is another question). Therfore, if the line is +8, the "side" to take (i.e. the one with relative value) would be the favorite, and if the published line is +14, the side to bet would be to take the points and the dog. <BR/><BR/>How long has this model been around? Is there enough data to statistically show a high confidence level that it is any better than flipping a coin? It seems sound, but would be curious if its been in action long enough to prove its worth.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-86356539549660742162008-11-18T16:54:00.000-05:002008-11-18T16:54:00.000-05:00To understand the expected winning margin using th...To understand the expected winning margin using the Win Probability one would need to track the winning margin for each WinProb, i.e.,0.83 wins by an average of 7.73 points.<BR/><BR/>For the year (2008), the Point Spread Favorite has won 111, lost 46, a percentage of 70.7, virtually identical to the 70.4% above. Again, won their games, not against the spread.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-21699694921230571112008-11-18T09:44:00.000-05:002008-11-18T09:44:00.000-05:00How can you determine the accuracy of this system ...How can you determine the accuracy of this system ATS? What you are giving here is a probability of a win. I can convert this to a M/L, but how do I know what "side" your model is predicting to win?<BR/><BR/>Good Stuff here!<BR/>ThanksAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-63443615955251554512008-11-14T17:27:00.000-05:002008-11-14T17:27:00.000-05:00Dan, right on. Would love to see the moneyline res...Dan, right on. Would love to see the moneyline results, taking all the teams as predicted. Small chore to figure that out though.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-78060813501448842072008-11-14T15:15:00.000-05:002008-11-14T15:15:00.000-05:00Dan, are you talking about the Kelly criterion?Dan, are you talking about the Kelly criterion?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-48529267293732209902008-11-14T12:31:00.000-05:002008-11-14T12:31:00.000-05:00You could also compare the information your win pr...You could also compare the information your win probabilities gives versus that in the Vegas money line. ie using your % causes the representation of results to use n bits, versus m bits if the Vegas line is used as a basis for generating the encodingAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-79596833287132978002008-11-14T12:29:00.000-05:002008-11-14T12:29:00.000-05:00It appears that you're working to maximize your ac...It appears that you're working to maximize your accuracy with respect to win probability. You could measure your relative predictive accuracy versus the money line.<BR/><BR/>If you consider a game to be 90% one way and the line is listing it as only 80%, you should bet about 82% of your bankroll on the favorite to optimize the growth of the log of your bankroll. You could further refine this by treating simultaneous games as a joint optimization problem as that's more realistic, but it's more complicated as you often don't have enough money to cover all the desired bets.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-26062813391446554702008-11-14T11:50:00.000-05:002008-11-14T11:50:00.000-05:00Keep in mind this is not intended to be an ATS sys...Keep in mind this is not intended to be an ATS system. I don't want to mislead anyone.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-31662812202552343052008-11-14T11:27:00.000-05:002008-11-14T11:27:00.000-05:00I've been keeping an eye on the ats record and it'...I've been keeping an eye on the ats record and it's currently 44 wins and 54 losses for 45%Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-63289101110962726202008-11-14T01:08:00.000-05:002008-11-14T01:08:00.000-05:00Do you by any chance know the system's ATS record?...Do you by any chance know the system's ATS record?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-64060540683033530232008-11-13T12:04:00.000-05:002008-11-13T12:04:00.000-05:00Yes. 69-29 (70.4%).Yes. 69-29 (70.4%).Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-9367036187566603532008-11-13T12:00:00.000-05:002008-11-13T12:00:00.000-05:00what is this system's straight-up record so far th...what is this system's straight-up record so far this season?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-70211107726134672412008-11-13T11:55:00.000-05:002008-11-13T11:55:00.000-05:00Do these probabilities take into account home fiel...Do these probabilities take into account home field advantage?Anonymousnoreply@blogger.com