tag:blogger.com,1999:blog-38600807.post8723159265097490840..comments2023-11-05T04:16:44.937-05:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Bayesian Draft Prediction ModelUnknownnoreply@blogger.comBlogger5125tag:blogger.com,1999:blog-38600807.post-2051952552060950902014-05-16T14:30:55.484-04:002014-05-16T14:30:55.484-04:00I've tried using Chrome, IE 8, and Firefox, bu...I've tried using Chrome, IE 8, and Firefox, but I can't see the inline graphs. The only graph that works is the final one.<br /><br />This analysis is so interesting I need to see the deets.Sean Shttps://www.blogger.com/profile/03902084229952883298noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-18527994265098788162014-05-08T13:51:06.377-04:002014-05-08T13:51:06.377-04:00So awesome. Well done sir!So awesome. Well done sir!Whamphttps://www.blogger.com/profile/16006695529627865508noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-85808307532329266472014-05-01T16:54:58.454-04:002014-05-01T16:54:58.454-04:00I'm posting my comment here instead of Twitter...I'm posting my comment here instead of Twitter. :)<br /><br />I built a similar model, but I fear I went down a wrong rabbit hole. When you were building this model, I noticed that you didn't account for a draftees on-field position. When I initially tackled this model, my "eye test" made me believe that the two biggest factors in a player "slipping" (i.e. where my model incorrectly had the projected draft position too high) were medical/off-field (i.e. not widely available knowledge) and the QB position. After reading your article, I feel like I'm going down the overfitting path. However, I feel like that might be my own biases and I'm in the process of examining some old mock draft data now to prove/disprove. <br /><br />TL;DR - Do you think such an analysis is even worth my time? Is there any value in potentially incorporating on-field position (i.e. QBs where there is a confined demand)?<br /><br />- @hagrinHagrinhttps://www.blogger.com/profile/05924051969292812536noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-68608085953195924992014-04-30T22:36:42.623-04:002014-04-30T22:36:42.623-04:00All the comments are on twitter these days...
Nat...All the comments are on twitter these days...<br /><br />Nat- Good question. The experts don't agree on most picks except in a few cases, so I trust they are using independent judgment. But there's no way for anyone to know for sure.<br /><br />Of course, they're all basing their analyses on the same factors, so obviously they'll never be independent in the strict probabilistic sense. But that's exactly what we're after. In this construction, probability is defined as 'degree of belief' rather than the frequency of an outcome.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-29320203901833009882014-04-30T14:12:15.660-04:002014-04-30T14:12:15.660-04:00Are you assuming independence of the projections y...Are you assuming independence of the projections you're using as predictors (essentially naive Bayes)? That seems like it might be creating over-confident estimates, though to know how bad the effect is you'd have to have some idea of how dependent the projections actually are.Natnoreply@blogger.com