tag:blogger.com,1999:blog-38600807.post7729050644324399726..comments2023-11-05T04:16:44.937-05:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Koko Fantasy Rankings - Running BacksUnknownnoreply@blogger.comBlogger6125tag:blogger.com,1999:blog-38600807.post-50737695927207893382009-08-26T09:52:15.879-04:002009-08-26T09:52:15.879-04:00i insist you use receptions/game. thxi insist you use receptions/game. thxRaHhttps://www.blogger.com/profile/14636068536257498024noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-57041265177485885502009-08-25T10:55:21.816-04:002009-08-25T10:55:21.816-04:00Brendan-We'll see! I agree RB is the toughest ...Brendan-We'll see! I agree RB is the toughest position for a naive system to project.<br /><br />Also, Chase raises a good point. I did the projections both ways, a straight year-to-year regression and a model regressing both per-carry stats and carries per game. <br /><br />The total projected points for each RB under both methods were very, very close (usually within about 3 points for the whole season), so I chose the simpler method (for now).Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-34353745574625958692009-08-25T06:50:49.685-04:002009-08-25T06:50:49.685-04:00Hm, it seems that Koko isn't so good with RB&#...Hm, it seems that Koko isn't so good with RB's. I don't think any of the projected top 5 RB' should actually be in the top 5, and most of them don't deserve to be in the top 10.Brendan Scolarinoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-65258490072716856622009-08-24T15:13:50.580-04:002009-08-24T15:13:50.580-04:00mvp,
I read the article -- I think you need to re...mvp,<br /><br />I read the article -- I think you need to regress both carries/game and YPC. YPC is highly susceptible to regression to the mean, as are carries per game. Rushing yards per game, of course, is, too. But I think in different ways. I think YPC is more susceptible to big plays, and therefore, players regress there at a quicker rate than carries per game.<br /><br />Maybe Brian can provide a more full explanation.Chasehttp://www.footballguys.comnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-88210816559219779452009-08-24T15:01:06.034-04:002009-08-24T15:01:06.034-04:00Chase - I suggest reading the article before you c...Chase - I suggest reading the article before you comment on it.<br /><br />IMHO, when you try a fit lines thru clouds of data, you need to keep in mind the scaling of your axes. The standard deviation of the data in Fig. 2 (rush yds/gm) looks to be about 20 yds/gm. So for a 67% confidence interval, your results should be within 20 yds/gm of the true number. That's a HUGE drop-off, especially when you factor in similar results for tds/gm. This is 1st round pick to 5th round pick type unpredictability.<br /><br />As always, when using statistics you can't just publish your results. You need to say how useful they are in making decisions. I think we'd both agree (after all, you call these the "koko fantasy rankings") that the answer is you probably shouldn't make ant decisions based on these data.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-55953926713691834022009-08-24T12:02:51.152-04:002009-08-24T12:02:51.152-04:00One thing I would do is break down rushing yards p...One thing I would do is break down rushing yards per game into YPC and carries. I think this might improve the results.Chasehttp://www.footballguys.comnoreply@blogger.com