Title: Interesting win probabilities from Ken Massey Post by: Clockwork Orange on October 03, 2011, 05:38:49 EDT Note that the schedule is in reverse and that any game not yet played has a win probability instead of a result . . . his system basically thinks that South Carolina and UGA are coin tosses but we are slight favorites at home; we have a chance at Arkansas on the road; we should chalk up Alabama and LSU as losses; Vandy, UK, and MTSU are highly likely to be wins. Add up expected values and Massey sees 4.4 more wins . . . so 7-5 or 8-4. That's probably a hair more optimistic than most Vol fans right now.
Code: Date Opponent Rk W L Result PF PA Title: Re: Interesting win probabilities from Ken Massey Post by: Clockwork Orange on October 03, 2011, 05:56:00 EDT Oh man, Massey's site can be fun. There's a little tool where you can see the probability distribution function for the margin, the total points, and each team's points. According to that, the median score for the UT-UGA game is 31-28, UT (as in, there's a 50% chance above and below those scores for each team). I believe he comes up with the distributions via repeated simulations.
Of note: * He gives us a 20% chance of scoring more than 40 points; UGA has a little under 15% chance of doing the same. * He gives us a 5% chance of holding UGA to two TDs or less; UGA has a 3% chance of doing this to us. * He gives us a 29% chance of winning by more than a TD, and UGA a 20% chance. Thus, there is a 51% chance that the game will be decided by a TD or less. Title: Re: Interesting win probabilities from Ken Massey Post by: Clockwork Orange on October 03, 2011, 06:04:58 EDT Okay, I know the above two posts were dorky and I may be the only one who enjoys that sort of thing . . . but this is a pretty cool function Massey puts on his site as well-- a schedule map:
http://masseyratings.com/map.php?s=107811&t=7801 (http://masseyratings.com/map.php?s=107811&t=7801) Title: Re: Interesting win probabilities from Ken Massey Post by: Clockwork Orange on October 03, 2011, 06:19:34 EDT I'll just continue talking to myself to highlight one more feature. I'll just call this the MTWHBYTB (My Team Would Have Beaten Your Team Because) function. I did last year for Tennessee beating Ohio State and got this:
Code: 2010-11-27 @ Tennessee 24 Kentucky 14 :laugh: Title: Re: Interesting win probabilities from Ken Massey Post by: PirateVOL on October 03, 2011, 06:31:52 EDT I'll just continue talking to myself to highlight one more feature. I'll just call this the MTWHBYTB (My Team Would Have Beaten Your Team Because) function. I did last year for Tennessee beating Ohio State and got this: You sir, require some serious help ... :beer:Code: 2010-11-27 @ Tennessee 24 Kentucky 14 :laugh: I'll check Massey's site when I get home this evening Title: Re: Interesting win probabilities from Ken Massey Post by: BanditVol on October 03, 2011, 07:21:55 EDT CO...that's very interesting. Because his calculations for USCe, uga and Arkansas match my subjective impressions of those games based on watching them all play other teams.
I guess I am a dork also, as I find this very interesting. :thumbup: |