Results 1 to 12 of 12
  1. #1
    Join Date
    Feb 2007
    Location
    Ashburn, VA

    Bracket scoring theory

    Usually I don't pay attention to the round weights, since I usually have a terrible bracket and it makes little difference to me. However, since I'm currently 3rd out of 60 in my office pool, I've been thinking a bit more about it. (13 out of 16 Sweet Sixteen correct, including NC State and Florida, and 3 out of 4 Final Four)

    It seems like a lot of sites go with 1-2-4-8-16-32 points for each of the "6" rounds. To me, that seems to overemphasize the later rounds, particularly the FF and NC games. I realize that it creates the same amount of possible points per round, but since a lot of this is guesswork and crap-shoots, and not statistically refined bets, that's concentrating a lot of that "luck" in one place, instead of distributed across multiple guesses, as in the first round (pause for breath).

    Do you think that's a good weighting scheme? Are there other major sites (I only checked 3) that do it differently? What about the following options:

    A) 1-2-4-8-16-32
    B) 1-2-4-8-12-16
    C) 1-2-4-8-10-12
    D) 1-2-4-6-8-10
    E) 1-2-3-4-5-6
    F) One of the above, but adding in a difference in seed for every upset. Example, for choice "D", a first round 5-12 upset would get 8 (1+7) points, while a S16 1-4 upset would get 7 (4+3) points. I only include this because that's how a friend runs his pool.

    To me, any seem better than "A".

    Currently a lady who only got 7 Sweet Sixteen teams correct, but also 3 out of 4 Final Four could shoot up from 24th out of 60 to 1st (!) if Ohio St. wins because almost no one else picked them to go all the way (as expected, tons of Kentucky, and a bunch of UNC and Kansas picks).

    So "A"'s downsides are only mitigated if 2 teams who a ton of people both picked to make the championship game actually do so - but that ends up being somewhat rare on the men's side.

  2. #2
    Join Date
    Mar 2008
    Location
    Mary's Place
    Agreed; I thought the doubling of each round placed too much emphasis on the final four. When I ran these "amusements" in previous years, I used this point system:

    1-2-4-8-12-24

    In your example, the person who went against the chalk and picked Ohio St should be rewarded; but those who picked 13 out of 16 shouldn't get wiped out either.

    The "underdog seed differential" is fun, but I think it's too easy to game - an actuary or finance guy can figure out expected risk (probable outcome) vs. reward (percentage who picked the underdog) and identify the upsets most likely to be net winners relative to most brackets...
    "Quality is not an option!"

  3. #3
    Join Date
    Feb 2007
    Location
    Austin, TX
    Quote Originally Posted by Turk View Post
    Agreed; I thought the doubling of each round placed too much emphasis on the final four. When I ran these "amusements" in previous years, I used this point system:

    1-2-4-8-12-24

    In your example, the person who went against the chalk and picked Ohio St should be rewarded; but those who picked 13 out of 16 shouldn't get wiped out either.

    The "underdog seed differential" is fun, but I think it's too easy to game - an actuary or finance guy can figure out expected risk (probable outcome) vs. reward (percentage who picked the underdog) and identify the upsets most likely to be net winners relative to most brackets...
    My favorite scoring method back when these were done manually was to give the points of the seed in the first round and 2x seed in the second round, then have flat scores throughout like below. Can't find a free website that offers this flexibility, though, so we've succumbed to the final four or bust ESPN pool.

    1: 1x Seed (144 to 400 total points available)
    2: 2x Seed (288 to 800 points available)
    3: 40 (320 points available) - key here -- any third round win is more valuable than a second round win, regardless of seed.
    4: 80 (320 points total)
    5. 120 (240 points total)
    6. 160 (160 points total)

    It seems as if the first 2 rounds are over-weighted, but if you pick rationally, your maximum points available in the first round will be MUCH closer to the minimum than the maximum. Picking upsets is great, but if you go overboard, you won't have anyone left.

    We typically also gave a 3x pool prize to the best score after round 1.

  4. #4
    Join Date
    Feb 2007
    Location
    St. Louis
    The pool I participate in awards point totals of 1, 2, 4, 6, 11, and 14 for the six rounds, adding the number of the team's seed in each game. No matter how you do it, there can be some unfairness. In our pool this year, the one person who correctly picked all four of the final four teams may or may not finish first or second.

  5. #5
    I like the Fibonacci System for scoring: 2, 3, 5, 8, 13, and 21 for the Championship.

    Essentially a win in any round is worth the equivalent of the two proceeding rounds. I think this tends to not completely overvalue the championship, while still maintaining enough of top-heavy weighting to still create drama during the Final Four matchups.

  6. #6
    Join Date
    Feb 2007
    Location
    Ashburn, VA
    For those of you with non-standard scoring schemes in your pools - do you use online sites that let you customize your scoring, or do you have to grade people's by hand still?

  7. #7
    Join Date
    Feb 2007
    Location
    St. Louis
    Quote Originally Posted by snowdenscold View Post
    For those of you with non-standard scoring schemes in your pools - do you use online sites that let you customize your scoring, or do you have to grade people's by hand still?
    CBS allows a limited, clumsy form of customization, that that's what we use.

  8. #8
    I am a grumpy curmudgeon who thinks that in a 30 person pool, at least 20 people should be all but eliminated after the first weekend. Fully 3/4 of the games are played by then, and in addition, 2/3 of those are games where everyone knows and can choose from both teams. I guess I just give a lot more credit for getting nearly all the first round games right then for happening to pick the eventual champ. Too much of a crapshoot. Doubling points in each round is ludicrous. Picking the title game should be nowhere near as valuable as a perfect Sweet 16.

    Understand the desire to keep interest around, though, and some escalation. I've pushed for something like Truth's proposal, with a -1 tweak when adding the two previous rounds, so 2,3,4,6,9,14. Folks complained that that still overemphasized the first two rounds. Anyone who picked both Norfolk State and Lehigh, for instance, should have something approaching an insurmountable lead unless the rest of their sheet is just horrendous. People shouldn't see so little chance for differentiation in the first couple rounds that they spend all their time just trying to pick the Final Four and make it a little bit different than their expectations of everyone else's Final Four so as to capitalize.

    So I've come to favor something like Truth's or Turk's, but with sufficient ability to capitalize on upset picks in the early rounds to make it worthwhile. I'd like to experiment with something like the Fibonnaci starting at a higher value: First Rd: 20+seed differential; Second Rd: 40 + (2x seed differential); and so on, with possibly diminishing factorizations on the upset bonus as raw value of the win increases.

    I think the ultimate system would figure out a way to properly value upsets based on their likelihood and how late in the tourney they occur.

  9. #9
    Join Date
    Feb 2007
    Location
    Skinker-DeBaliviere, Saint Louis
    I guess this hasn't been done in a while, but back in the 1990s, I was in several pools where you simply picked every game as it happened, and whoever's record was best was the winner. In 1994, me and my HS BB coach were tied at like 42-20, and he took Arkansas in the final and I took Duke and he won.

    I supposed people don't do that anymore. The advantage it offers is that it's like September in a pennant race. You can have three or four people within a couple games of each other. It's not all about whichever crazy nut picked UConn to win it all in 2011. Or who got insanely luck on Norfik State.

    A movie is not about what it's about; it's about how it's about it.
    ---Roger Ebert


    Some questions cannot be answered
    Who’s gonna bury who
    We need a love like Johnny, Johnny and June
    ---Over the Rhine

  10. #10
    Join Date
    Jun 2010
    Location
    Roxboro, NC
    I have been doing the office pool here at work for several years now and we use the CBS site. I probably don't have enough weight on the later rounds but no one has complained yet. I use 2, 3, 4, 5, 6, and 8 for the 6 rounds. I don't like how espn scores it. On Mike and Mike last week they said that Golic was in the 95th percentile and Greenie was 30ish, but Greenie could still beat Golic because of like 2 different games in the Elite 8 or something. That seems crazy to me. With their scoring the first 2 rounds are irrelevant.

    We usually have 20-30 entries in our pool and by the Final Four weekend our winners (1st and 2nd place) are narrowed done to 2-7 people. This year the top 2 are already locked in but they can change spots based on the KU/OSU game. That is mainly because most everyone picked either UK or UNC to win so no one can make up any ground. I picked 3 of the 4 Final Four teams (missed UK for some reason) but I'm close to the bottom in the standings because I did terrible in the first 2 rounds.

  11. #11
    I love how the complainers of a heavily weighted back-end make the false claim that the first two rounds are irrelevant.

    They aren't irrelevant. If you pick incorrectly, you don't have a chance for success because you have no chance of scoring in the later rounds.

    Is it sad if somebody gets 13 out of 16 SS teams right, but then tanks every regional semi, or misses on the whole final 4, so loses? Maybe, but I don't buy it.

    That's why you score the whole bracket. I put more credence in being able to select the team that wins 6 games than in 28 of the 32 first round games. That showed the ability to look forward and think critically about what-if situations.

    I will say the Fibonacci scoring is interesting to think about, with a win in each round being worth the sum of wins from previous 2 rounds. Of course, that limits the scoring in later rounds, too - maybe too much.

    64 points available in Round 1
    48 in round 2
    40 in regional semis
    32 in regional finals
    26 in the FF semis
    21 in the final game

  12. #12
    Join Date
    Jun 2010
    Location
    Roxboro, NC
    Quote Originally Posted by cf-62 View Post
    I love how the complainers of a heavily weighted back-end make the false claim that the first two rounds are irrelevant.

    They aren't irrelevant. If you pick incorrectly, you don't have a chance for success because you have no chance of scoring in the later rounds.

    Is it sad if somebody gets 13 out of 16 SS teams right, but then tanks every regional semi, or misses on the whole final 4, so loses? Maybe, but I don't buy it.

    That's why you score the whole bracket. I put more credence in being able to select the team that wins 6 games than in 28 of the 32 first round games. That showed the ability to look forward and think critically about what-if situations.
    You are right for the most part, but as long as you advance most of your final four, then the rest is irrelevant. For example; Let's say in this year's bracket you had a perfect first 2 rounds and got 640 points using espn's scoring. (32*10, then 16*20) Meanwhile in my bracket I got everything wrong in the first 2 rounds except for 8 teams including my final 4, for a total of 240 points. (8*10, then 8*20) So you picked 48 games right to my 16. But if you picked UNC and I picked UK to win it all, I would still beat you by a substantail margin if UK wins. Even if we both picked all 4 final 4 teams, if we have differential in the Championship game then I can still beat you. That just seems absurd to me.

Similar Threads

  1. I can't stand The Big Bang Theory
    By Jim3k in forum Off Topic
    Replies: 49
    Last Post: 10-18-2011, 05:45 PM
  2. Lost - Time Loop Theory
    By Dukerati in forum Off Topic
    Replies: 6
    Last Post: 03-28-2008, 01:48 PM
  3. Taylor King Theory
    By houstondukie in forum Elizabeth King Forum
    Replies: 42
    Last Post: 03-24-2008, 07:11 PM
  4. Lost theory
    By JasonEvans in forum Off Topic
    Replies: 1
    Last Post: 03-16-2007, 12:41 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •