Welcome home, fellow Gator.

The Gator Nation's oldest and most active insider community
Join today!

Playoff Chances according to the pundits...

Discussion in 'RayGator's Swamp Gas' started by ArtVandelay, Oct 10, 2019.

  1. your_perfect_enemy

    your_perfect_enemy GC Hall of Fame

    5,543
    378
    298
    Apr 7, 2010
    What's the point of playing then? We beat them soundly and if we played them 99 more times would beat them in at least 79 of those. They flat out aren't a better team than we are and it was proven last Saturday.

    Their own stupid system even has them finishing with a worse record
     
    Last edited: Oct 10, 2019
  2. GatorRade

    GatorRade Rad Scientist Premium Member

    7,072
    704
    708
    Apr 3, 2007
    Should Ole Miss have been ranked ahead of Florida in 2008? Single outcomes are just that. What happens if we upset by Vandy later? Should Vandy be ranked ahead of us and Auburn?

    Remember the point is to be a predictive model, not a resume ranking (though Auburn actually two good wins to our one, so they might even have an argument there) Basically, deserve’s got nothing to do with it.

    Of course they do. In addition to LSU and Georgia, they play Bama. You don’t think strength of schedule should have anything to do with record projections?
     
  3. your_perfect_enemy

    your_perfect_enemy GC Hall of Fame

    5,543
    378
    298
    Apr 7, 2010
    If they had the same or better record than us, then yes they should have been ranked ahead of us at that point.

    Of course strength of schedule matters, but they have ours through the rest of the year as the toughest but have us with a better record.
     
  4. GatorRade

    GatorRade Rad Scientist Premium Member

    7,072
    704
    708
    Apr 3, 2007
    The point of the metric is to be predictive only. No model worth anything would have predicted Ole Miss to be the more powerful team in 2008.

    When all you look at is total record and one head to head outcome, you are ignoring a load of other informative data about quality of record and margins of victory. Therefore, the team with the best resume isn’t always the most powerful team.

    For example:

    USC 'upset' No. 3 Utah, just like stats predicted

    The record projection that you are seeing is for the entire season. The SOS projection is for the back half of the season only. If you look at the FPI win projection for the back half of the season only, you’ll see that they actually predict a slightly better record for Auburn.
     
  5. WhattaGator

    WhattaGator Ever Vigilant Psycho Mod Moderator VIP Member "Cook Shack Chef"

    So....

    Our remaining SOS is ranked #1

    When we beat LSWho on Saturday, how much more respect are the pundits going to give us??
     
  6. your_perfect_enemy

    your_perfect_enemy GC Hall of Fame

    5,543
    378
    298
    Apr 7, 2010
    That was my original point, predicatively it’s wrong. I don't care what their little data says, If we played auburn a bunch of more times we’d beat them many more times than not.
     
  7. merinker

    merinker Junior

    158
    37
    78
    May 28, 2019
    78% at what? Picking games straight up? That doesn’t seem very impressive to me at this point in the season given the huge mismatches with so many FCS teams on schedules so far.
     
  8. merinker

    merinker Junior

    158
    37
    78
    May 28, 2019
    The “betting line” has nothing to do with ranking teams. It only has to do with generating equal action on both sides of a game or on all sides of futures.
     
  9. sierragator

    sierragator GC Hall of Fame

    7,860
    331
    333
    Apr 8, 2007
    right now it is an exclusive club with bama, Clemson, uga, an ahia state university, ok, with a few outliers like LSU.( and of course the eternal darling of ND).. teams like us aren't on the "guest list". Screw the cover charge, cut the rope and bust the damn door down
     
    • Like Like x 1
  10. GatorRade

    GatorRade Rad Scientist Premium Member

    7,072
    704
    708
    Apr 3, 2007
    That’s your opinion based on your eyes. Maybe you’re right, but it hasn’t happened and to date, no single person’s eyes have been consistently better than Vegas predictions, so maybe not.

    Plus, your head to head model quickly runs into trouble. What if Oregon beats Florida? Then you rank Florida above Auburn, Auburn above Oregon, and Oregon above Florida?

    Using a full season’s worth of data is always going to be more predictive than using single outcomes.
     
  11. GatorRade

    GatorRade Rad Scientist Premium Member

    7,072
    704
    708
    Apr 3, 2007
    I understand what you’re saying, but I would suggest that these two statements are at odds. People would only bet on LSU to win by 13+ if they thought that LSU would win by 13+. And since betting is open to everyone, the line ends up being a great information aggregator. And backed with confidence, as real monetary wages are at stake. It is a prediction market.

    And regardless of whether it’s intending to predict an outcome, the betting line is a very accurate predictor of outcomes relative to competing instruments (see the link above). I did an analysis of 10 years of college football betting lines and found that the size of the line is very well correlated with win percentage (r square = 0.96). That analysis showed that teams favored by 13 won the game about 80% of the time.

    So yes, of course the line “only” balances the for and against bets, but that’s in itself is an extremely powerful predictive tool.
     
  12. Distant Gator

    Distant Gator GC Hall of Fame

    4,653
    553
    393
    Apr 9, 2007
    Upstate, SC
    Rade- per the link you posted, even the best of those predictive models barely went over 50% ATS.
    So even they, with their vast resources, would not make money once the 10% vig is taken into account.

    That tells the spread is really accurate- which I have seen as I play the "picks panel" game on GC.
    Many times, the spread is covered or broken based on a total fluke. (Like the UT fumble return for a TD by UGA this past weekend. It's how UGA covered.)

    So IMO those spreads are really accurate. But obviously any one game be a complete outlier. (Like when OM beat us in 08.)
     
    • Winner Winner x 1
  13. GatorRade

    GatorRade Rad Scientist Premium Member

    7,072
    704
    708
    Apr 3, 2007
    Absolutely. Great point. And even the ones that are above 50% right now probably wouldn’t do it again next year. Vegas is about as good as it gets.
     
  14. Distant Gator

    Distant Gator GC Hall of Fame

    4,653
    553
    393
    Apr 9, 2007
    Upstate, SC
    As an aside, I had 3 games this year be decided by total flukes during the last minute.
    USCe vs Bama. USCe scored a TD w/ 11 seconds left to cover.
    TAMU vs Clemson. TAMU scored a TD w/ under 10 seconds to cover.
    UGA vs UT. Already mentioned, but fumble returned for TD.

    All these TDs were meaningless within the game, but were basically flukes. I mean USCe and TAMU could have kneeled it out to finish the game.

    The Vegas lines are amazingly accurate. A fluke either way seems to decide them. Makes it tough to bet them. (Glad we don't use real $)
     
    • Fistbump/Thanks! Fistbump/Thanks! x 1
  15. reaves2alvarez

    reaves2alvarez Senior

    225
    151
    308
    Nov 29, 2016
    We can lose one game as long as we make it to the SECÇ and win it. There will be enough there to make top 4. Beautiful thing about our league. You control your own destiny every year.

    “Wes Chandler was Percy Harvin before Percy Harvin was born.”
     
    • Agree Agree x 1
  16. Gatorrick22

    Gatorrick22 GC Hall of Fame

    64,980
    10,253
    2,373
    Apr 3, 2007
    I always thought that the quality of top 25 wins was more important than winning against some inferior teams.