Help with game prediction contests

Discussion in 'On The Field' started by Dionysus, Aug 13, 2019.

  1. Dionysus

    Dionysus Idoit Admin

    Last year I thought it was a lot of fun when @LonghornCatholic would post a prediction thread a few days before a game. Everyone would post their score, then it would all go into a table like the one below (Sugar Bowl) that gets posted right at game time, with winners acknowledged afterwards. Winner (or co-champs) gets a little banner thingy on their avatar which we will leave in place through the entire season. HF bling.

    Here’s what I’d like help with.

    So obviously if someone predicts the exact outcome, that’s the winner. Or co-champs if more than one person picks the final score correctly. End of contest.

    But what if nobody gets the right final score? What we did last year was whoever got the spread right would be the winner(s) in this case. Is that the best way to do it? Or what about whoever got closest to the final score — what should the rules of the contest be? Let’s figure something out.



    HFTEXASGEORGIASPREAD
    4th_floor4945Texas by 4
    AC3728Texas by 9
    BabHorn3831Texas by 7
    Bill in Sinton3120Texas by 11
    blonthang3331Texas by 2
    bystander2831Georgia by 3
    caryhorn1727Georgia by 10
    CraigHorn3835Texas by 3
    Creek2435Georgia by 11
    Crockett3734Texas by 3
    Dionysus3026Texas by 4
    Driver 83530Texas by 5
    dukesteer3424Texas by 10
    EDT2724Texas by 3
    Garmel2428Georgia by 4
    Giovanni Jones2428Georgia by 4
    Hoop2731Georgia by 4
    Horn2RunAgain2048Georgia by 28
    Htown773431Texas by 3
    HTownBevo3721Texas by 16
    humahuma2435Georgia by 11
    IvanDiabloHorn4341Texas by 2
    joh3531Texas by 4
    LongestHorn5141Texas by 10
    LonghornCatholic3027Texas by 3
    LonghornDave3431Texas by 3
    militaryhorn2031Georgia by 11
    moondog_LFZ4532Texas by 13
    Pomspoms2824Texas by 4
    Run Pincher2341Georgia by 18
    ShAArk923324Texas by 9
    stanhin2028Georgia by 8
    Statalyzer2034Georgia by 14
    Texex81109Texas by 1
    theiioftx2827Texas by 1
    ViperHorn3328Texas by 5
    Vol Horn 4 Life3138Georgia by 7
    X Misn Tx3431Texas by 3
     
    • Like Like x 1
  2. Sangre Naranjada

    Sangre Naranjada 10,000+ Posts

    I think if you combine the difference between
    1) Absolute value of the difference between Texas' predicted score and reality
    2) Absolute value of the difference between opponents' predicted score and reality
    2) Value of the difference between Spread predicted and Spread reality
    You'll arrive at a good place.

    Example
    Contestant A
    Texas 28
    Victim 21

    Contestant B
    Texas 35
    Victim 28

    Contestant C
    Victim 31
    Texas 27

    Actual score
    Texas 31
    Victim 27
    _________________
    Contestant A is calculated as 3+6+3 (Texas, opponent, spread)
    Contestant B is calculated as 4+1+3
    Contestant C is calculated as 4+4+8 (4 pt spread was correct but in the wrong direction, or think of it as predicting Texas to win by negative 4 points)

    Low overall score wins
    _________________
     
    • Like Like x 2
  3. Vol Horn 4 Life

    Vol Horn 4 Life Good Bye To All The Rest!

    I think they have to have picked the correct game winner and closest point total or closest point spread.
     
  4. Dionysus

    Dionysus Idoit Admin

    I need this to be easy so I think what I want is something like this.

    Winner is whoever:
    1. Picks the final score correctly, or
    2. Picks the spread correctly, or
    3. Picks total points correctly and has the right winner

    I’m open to other suggestions but just want something I can figure out quickly and post results after the game.
     
    • Like Like x 1
  5. LousianaHorn

    LousianaHorn Kabong

    I vote for #1 Dion, isn't that the way it was done last year?
     
  6. Ajo Macho

    Ajo Macho 500+ Posts

    1. Predicts final score. Ties possible.

    If no one predicts final score
    2a. Predicts spread. Ties likely.
    2b. Tiebreaker: Closest to total points. Ties still possible.

    Example: prediction was 31-21, score was 30-20. Correctly predicted spread, was off by 2 total points (predicted 52, was 50). Better than predicting 45-35 (80 total points, off by 30).

    If no one predicts spread
    3a. Closest to predicting spread. Ties likely.
    3b. Tiebreaker: Closest to total points. Ties still possible.
     
  7. X Misn Tx

    X Misn Tx 2,500+ Posts

    I think that #1 is the best.

    However, I'm not a big fan of someone picking a game 14-10 and winning because the final score was 38-34. The actual score has to have some reasonable impact.

    But it's all in fun.

    Picks closest to the actual winning score chosen by margin?
     
    • Like Like x 2
  8. Dionysus

    Dionysus Idoit Admin

    Yes, and we need an alternative if nobody picks the final score, so in that case we would go to #2 for a winner, and if nobody meets that criterion, go to #3.

    Your 2b is basically what I envisioned for #3. Total points scored or closest to that, but I think you should have the correct winner in this case. Ties are fine because we can have co-champs for each game.
     
  9. AC

    AC 2,500+ Posts

    #1 is my vote. Thanks Dion.
    :bevo:
     
  10. Dionysus

    Dionysus Idoit Admin

    Good, and just to be clear... I have a list of three items because if #1 is not met (nobody guesses the final score), then we go to #2 for a winner (spread), and if nobody guesses that then we go to #3.
     
    • Agree Agree x 1
  11. AC

    AC 2,500+ Posts

    Copy that. Could 2b be closest to final score? Like if the final is 31-10 and somebody guessed 30-10 and they were the closest to the actual score. But somebody else guessed 41-0 I would think 30-10 wins over 41-0. Just my $.02.
     
  12. dukesteer

    dukesteer 5,000+ Posts

    I’m hoping that Badshortgame chimes in. He had a good idea last year but I forgot what he proposed.

    To me, the spread is irrelevant if the total points are out of wack.
     
  13. Dionysus

    Dionysus Idoit Admin

    I looked for his post about this but didn’t find anything—could it have been another poster?
    I tend to agree but we just need some data points for awarding winners, which I think makes it more fun if we can find at least one winner every week.
     
  14. Statalyzer

    Statalyzer 10,000+ Posts

    Total combined points isn't good if that's what we're talking about.

    If the actual score was 21-20, then someone who predicted 38-6 got the total points closer than did someone who predicted 20-17, but the latter person clearly did a much better job of predicting the game.
     
    • Agree Agree x 1
  15. Galvestonhorn

    Galvestonhorn 250+ Posts

    lst: Did your pick win the game
    2nd: closet to point spread
    3rd: closest prediction to actual score of winning team.

    The point spread is what you absolutely have to consider first.
     
    • Agree Agree x 2
  16. Dionysus

    Dionysus Idoit Admin

    Y’all are trying to make me do calculus here I just want an easy little prediction game.
     
  17. 4th_floor

    4th_floor Dude, where's my laptop?

    I like what Sangre Naranjada proposes. I think if you add a disqualification for picking the wrong winner, the SN proposal is best. For example, if Texas beats LSU 3-0, and the Oscarthegrouch picks LSU to win 3-0 to give a total score of 3+3+6=12, and OptimistHorn is second lowest score as he picks Texas to win 31-17 for a score of 28+17+11=56, OptimistHorn wins and OscartheGrouch is taken outside and flogged for picking LSU.
     
    • Like Like x 2
  18. AC

    AC 2,500+ Posts

    It appears Galvestonhorn, Statalyzer and I are pretty much in agreement. We need a consensus so as not to make this more difficult. Less math and a more obvious path to picking the winner. The KISS principle. Hope that helps.
     
  19. Statalyzer

    Statalyzer 10,000+ Posts

    This has the opposite version of the same problem that total combined points over spread has: You predict 51-48 and the actual score is 16-13. Yeah, you got the winner and point spread right, but you were still way off compared to the guy who said 17-10.

    Since both point spread and total points matter, you almost have to have some sort of math that combines the two categories.
     
  20. AC

    AC 2,500+ Posts

    Stat,
    I think if 3rd moved to 2nd we may be closer.
    1. Pick the winner of the game.
    2. Pick closest to the actual score of the game.
    3. Point Spread (this won't matter much if 1 and 2 are picked).
    I was saying the same thing as you are. The winner and closest to final score are the highest priorities. Correct?
     
  21. Statalyzer

    Statalyzer 10,000+ Posts

    I guess it depends on how you define "closest to actual score". See my earlier comment:

    "If the actual score was 21-20, then someone who predicted 38-6 got the total points closer than did someone who predicted 20-17, but the latter person clearly did a much better job of predicting the game."
     
  22. militaryhorn

    militaryhorn Prediction Contest Manager

    It's not total points. It's each team's final score compared to each team's score predicted. In your scenario based on this parameter:

    1. Correct Winner (both in this case)
    2. Closest to actual score (20-17 compared to 38-6 when final was 21-20; 20-17 clearly is the best guess)
    3. Spread (not needed based on number 2)

    Number 3 only comes into play if you have 24-20, 23-20, 21-19 type of guesses so close to each other.
     
    Last edited: Aug 16, 2019
  23. nashhorn

    nashhorn 5,000+ Posts

    I’m with military:
    1. Pick winner
    2. Subtract prediction from actual (whole number) of each team. Lowest # wins.
    3. Then if needed, total points scored.
     
  24. theiioftx

    theiioftx Sponsor Deputy

    I’d go with #1 with a tie breaker being the spread.
     
  25. dukesteer

    dukesteer 5,000+ Posts

    Apologies to anyone who has already proposed basically the same thing.

    After picking the correct winning team, because using the point spread without factoring in the total points predicted lacks perspective, and because using the point total without factoring in the point spread also lacks perspective, I would suggest that both the point spread and the total points are used, together, but weighted differently. I would weigh the point spread at 65% and the point total at 35%.

    Anyone who picks both the actual point spread and the total points would have a deviation of 0, and they would win unless another individual also has a differential of 0, in which case there would be co-winners. Of course, that scenario is unlikely.

    So the winner each week would be the prognosticator whose variance from the actual total — point spread differential + total points differential — is the lowest, as factored at 65/35.

    This calculation appears to be complicated but in reality it is not. I believe it to be the most equitable approach.

    Example:

    A picks 48-47.
    B picks 21-7
    Actual score is 24-23.

    “A” was way off but he/she picked the spread correctly.

    “B” was much closer to the actual score but missed the spread big-time.

    “A” should be rewarded for correctly picking a close game, but A was way off on the point total. For that reason I like the combined, weighted approach.
     
  26. Dionysus

    Dionysus Idoit Admin

    I like the ideas proposed but here’s the thing.

    After each game, each week, I need to figure out the winners(s) and post it to the thread. I don’t want to spend a bunch of time crunching numbers. I need a fairly quick and simple way to determine winners.

    1. Final score — this is easy to see by just looking at the picks in the table. The data is there.
    2. Spread — same, this is captured in the table.
    3. Total points — I could add a column that calculates this if we need it.

    Maybe I need to add some columns to calculate a couple other data points, but whatever we do I want to have all the data in the table so I can get it at a glance.

    Or, if there is a spreadsheet guru here who enjoys spending time on the analysis I’m happy to let them generate results every week. Anybody?
     
  27. Texanne

    Texanne 5,000+ Posts

    If no one predicts the final score correctly, just award me as the winner.

    Problem solved.
     
    • Winner Winner x 2
    • Agree Agree x 1
    • Hot Hot x 1
  28. dukesteer

    dukesteer 5,000+ Posts

    Dion, I just sent you an email to Admin@ that included a comprehensive spreadsheet that will populate the winner automatically based on reasonable stipulations.

    Take a look and let me know. (I can’t take credit for the Excel spreadsheet.)
     
    • Like Like x 1
  29. Dionysus

    Dionysus Idoit Admin

    Got it and will try to digest it within the next couple of days. Hopefully we can figure out something that will be narrow enough to fit within a post here, so may need to minimize columns. Glad to have something to work with though!
     
  30. dukesteer

    dukesteer 5,000+ Posts

    Sent a second one that still weighs point differential at 70% but only (also) uses “winning team” points variance at 30%. This eliminates some inconsistency nuances.

    Regarding posting, the calculation cells can be made invisible such that only the scores and pics are shown.
     
    • Like Like x 1

Share This Page