The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Media is overstating how hard it is to win the Billion Bracket
Poast new message in this thread
Date: March 18th, 2014 7:54 AM Author: rambunctious indian lodge gunner http://fivethirtyeight.com/features/buffetts-billion-wont-lead-to-a-perfect-bracket/
"To our knowledge, no one has ever produced a perfect bracket, the three-minute mile of office pools. Perhaps one is buried in the boneyard of photocopied, hand-filled brackets of yesteryear. Internet-era contests are easier to track. Against the standard of perfection, more than 30 million ESPN brackets have failed, one by one, in the past 16 years, according to a spokeswoman. Over the last two years, no CBSSports.com bracket remained perfect through the second day of the tournament, a spokeswoman said. Yahoo’s entrant who came closest to perfection got 58 of 63 games right in 2007, according to a spokesman; last year no one picked more than 50 games correctly at either CBS or Yahoo. (Yahoo is partnering with Buffett’s perfect bracket contest.)"
"2) No one will win.
ESPN has been running a large-scale bracket contest for 13 years. Nobody has ever come close to perfection, the sports network’s John Diver told CNN in January. Only one person in the last seven years managed to pick just the first-round winners correctly.
“I don’t want to say it’s impossible,” Diver said, “but it’s basically impossible.”"
----------------------------------------------------------
ok look, its obviously a ridiculous long shot, but those articles are overstating it,
tradition brackets aren't a good predictor of how this bracket will play out because the incentives are different
in most polls its sort of silly to shoot for perfection for 2 reasons
1) if you're playing against other people, missing big upsets don't hurt you, if no one picked the #15 seed to win, the #15 seed winning doesn't hurt you relatively,
2) in most bracket, later rounds are seeded more heavily, so if your picking a big upset and miss, you're leaving alot of potential future points on the table
both of these reason disincentive picking alot of upsets
neither reason applies here, in fact in order to put in a plausible bracket, you better roll the dice on some fairly unlikely bracket
from my counting yesterday, if you want history to be your guide, you should probably pick 8-10 9 or lower seeds winning in the first round, and 3-5 12 or lower seeds winning in the first round
which would be way more agressive than would make sense in a traditional bracket
|
Date: March 18th, 2014 9:43 AM Author: rambunctious indian lodge gunner (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212237) |
|
Date: March 18th, 2014 9:47 AM Author: electric crawly puppy The incentives are different, but you're still taking what amounts to a random guess on at least some of the upsets, which puts you back at the statistical odds, which are really, really low. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212266) |
 |
Date: March 18th, 2014 9:54 AM Author: rambunctious indian lodge gunner of course
but not as low as "OMG no one's ever done it, its impossible"
its never been tried on a mass scale
picking 14 and 15 seeds is usually reserved for the idiots |
|
Date: March 18th, 2014 9:55 AM Author: painfully honest reading party Lol, it's 1 in 9 quintillion. Good luck getting to scale. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212315) |
|
Date: March 18th, 2014 9:58 AM Author: rambunctious indian lodge gunner no its not
the range of plausibly random outcomes is way smaller
also Buffet has already said he'll buy someone out if they make it to the final four
if they run it for 10+ years, someone will make it to the final four with a shot |
|
Date: March 18th, 2014 10:00 AM Author: painfully honest reading party Jeff Bergen, a math professor at DePaul University, derived a more realistic calculation that takes basketball knowledge into account. If you know the sport pretty well, he concludes, your chances of picking perfectly are more like 1 in 128 billion.
Still not so hot. As Bergen explained, that would mean you’d need to fill out about 90 billion brackets before you even had a 50-50 chance to win. |
|
Date: March 18th, 2014 10:02 AM Author: rambunctious indian lodge gunner yeah, that's about the odds I would put someone intellegently trying to be perfect at
and your odds are better if you're just trying to make it to the point where you can get bought out |
|
Date: March 18th, 2014 10:00 AM Author: painfully honest reading party Okay, yeah, that's not bad. I'd definitely let Warren Buffet buy me out. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212349) |
|
Date: March 18th, 2014 10:04 AM Author: rambunctious indian lodge gunner take it to Wall Street, let hedge fund managers bid on the buy out price
(probably in the fine print that Buffett's the only one who can make a buyout bid) |
|
Date: March 18th, 2014 9:58 AM Author: painfully honest reading party If all 317 million people in the U.S. filled out a bracket at random, you could run the contest for 290 million years, and there’d still be a 99 percent chance that no one had ever won. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212331) |
|
Date: March 18th, 2014 10:00 AM Author: rambunctious indian lodge gunner why would you fill your bracket out at random? (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212346) |
|
Date: March 20th, 2014 11:35 PM Author: Wonderful maniacal parlor wagecucks I don't know, just to be random (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25230558) |
|
Date: March 18th, 2014 10:00 AM Author: painfully honest reading party Jeff Bergen, a math professor at DePaul University, derived a more realistic calculation that takes basketball knowledge into account. If you know the sport pretty well, he concludes, your chances of picking perfectly are more like 1 in 128 billion.
Still not so hot. As Bergen explained, that would mean you’d need to fill out about 90 billion brackets before you even had a 50-50 chance to win. |
|
Date: March 18th, 2014 10:14 AM Author: stimulating aquamarine dingle berry ticket booth Lol at trusting a depaul bro. Get a loyola scholar in here and we'll see. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212443) |
|
Date: March 18th, 2014 10:01 AM Author: Magenta underhanded associate location (1/2)^63 (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212353) |
|
Date: March 18th, 2014 10:02 AM Author: rambunctious indian lodge gunner why would you do it at random? (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212357) |
|
Date: March 18th, 2014 10:03 AM Author: Magenta underhanded associate location you either win or lose 50/50 (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212365) |
|
Date: March 18th, 2014 10:09 AM Author: citrine senate faggot firefighter cr, can we put the 1/9quintillion thing to rest? Those are the odds of picking 63 50/50 coin flips correct, not weighted matchups. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212406) |
|
Date: March 18th, 2014 10:20 AM Author: stimulating aquamarine dingle berry ticket booth Brother each game is essentially a coin flip because anything can happen in March Madness ... if the slipper fits ... for One Shining Moment ... (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212496) |
|
Date: March 18th, 2014 10:41 AM Author: rambunctious indian lodge gunner lol, no its not (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212656) |
|
Date: March 18th, 2014 10:47 AM Author: painfully honest reading party Are you unfamiliar with the law of 50/50? (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212671) |
|
Date: March 18th, 2014 10:43 AM Author: rambunctious indian lodge gunner yeah
how many random combinations are there if posit that we have to have 8-10 >=9 seeds winning in the first round, and 3-5 >=12 seeds winning in the first round, and 0 16 seeds winning, and we don't give a shit what happens in the final four because we going to take the buyout number
any math people want to take a shot at how many of the 1/9quintillion brackets we have left? |
|
Date: March 18th, 2014 10:47 AM Author: painfully honest reading party 1/quintillion (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212676) |
|
Date: March 18th, 2014 10:50 AM Author: rambunctious indian lodge gunner lol (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212688) |
|
Date: March 18th, 2014 10:21 AM Author: clear marvelous candlestick maker But even sick knowledge won't win it. You need to be lucky to the tune of needing to be like, 1 in 150,000,000,000. It's hard enough to get out of the first round perfect, so hard that nerds have done it once in like 16 years. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212500) |
|
Date: March 18th, 2014 10:46 AM Author: rambunctious indian lodge gunner nerds don't try to get the first round perfect, they preserve the value of the rest of the bracket (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25212670) |
|
Date: March 19th, 2014 10:49 AM Author: citrine senate faggot firefighter IIRC, every years there's usually a good number of brackets that come out of the first round perfect on the ESPN challenge. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25219106) |
|
Date: March 19th, 2014 9:22 AM Author: rambunctious indian lodge gunner Nate Silver thinks he has it down to 1/7 billion
http://fivethirtyeight.com/datalab/we-have-a-1-in-6001225228-chance-of-winning-buffetts-billion/
which still seems like the wrong way to do it
if you have to pick 10 different game all with 60% chance of winning,
are you odds of picking every single game better picking all 10 favorites
or 6/10 favorites?
seems like expecting 10/10 favorites to win is obviously the wrong way to do it
(though there are more random combinations of 6/10 to sort through) |
|
Date: March 19th, 2014 10:51 AM Author: Dull big step-uncle's house For any one bracket, your odds are obviously better picking all 10 favorites. But when there are multiple brackets involved, its obviously important to ensure that each bracket is unique. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25219113) |
|
Date: March 19th, 2014 11:11 AM Author: rambunctious indian lodge gunner isn't there a distribution of how many favorites should win?
wouldn't 10/10 favorites winning be outside that distribution? |
|
Date: March 20th, 2014 8:51 AM Author: Submissive vigorous parlour Even if the most likely aggregate outcome is 6 favorites and 4 underdogs, the most likely individual bracket is still 10/10 favorites. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225146) |
|
Date: March 20th, 2014 8:55 AM Author: rambunctious indian lodge gunner so if this happens at all, does it almost have to happen in a historically low number of upsets year
or should that be the bet that makes the most sense to make? |
|
Date: March 20th, 2014 9:07 AM Author: Submissive vigorous parlour I think it's less total number of upsets and more that you just can't have teams like Florida Gulf Coast last year or Cornell a few years ago at all. Both of those teams probably eliminated 99% of brackets from contention to be perfect, and all of the sudden the "well maybe someone hits" becomes "impossible" again, esp. because it will probably eliminate an even larger portion of the "knowledgeable about basketball" brackets, meaning the brackets left go back up to the already unreasonable odds. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225174) |
|
Date: March 20th, 2014 9:12 AM Author: rambunctious indian lodge gunner considering the historical average is for 1 or so 14 or 15 seed to win
what do you think of the strategy of using Sagrin, (or someother rating system) to guess which is the most likely 2-3 14-15 upsets are, and then picking one of those? |
|
Date: March 20th, 2014 9:28 AM Author: Submissive vigorous parlour It's fine, and probably close to what most people actually do. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225222) |
|
Date: March 20th, 2014 9:29 AM Author: disrespectful alpha pervert lol @ "down to"
people have no idea how much 7 billion is. |
|
Date: March 19th, 2014 10:46 AM Author: rambunctious indian lodge gunner (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25219093) |
|
Date: March 20th, 2014 8:47 AM Author: rambunctious indian lodge gunner http://tournament.fantasysports.yahoo.com/quickenloansbracket/492066
rate my methodology
when with historical averages for number of upsets: 1 14-15 seed; 3 12-13 seeds; 5 9-11 seeds
then there after roughly 20-30 % of games with the lower seed winning
then used the Sagrins ratings to try and figure out which upsets in those categories were most likely
|
|
Date: March 20th, 2014 9:15 AM Author: obsidian exciting nursing home what if instead of a single entry, you got 7,456 unique entries? what are the odds then?
http://www.takebuffettsbillion.com |
|
Date: March 20th, 2014 9:18 AM Author: drab forum Yeah, SUMMON: that UCLA dood who gamed OkCupid
But srsly, devise something that creates a unique bracket on each entry. Could it be all that hard? |
|
Date: March 20th, 2014 9:23 AM Author: obsidian exciting nursing home they not only did that, but they generated algorithmically the most likely brackets, so those 7,553 entries are (more or less) the 7,553 most likely outcomes, not just random selections. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225208) |
|
Date: March 20th, 2014 9:25 AM Author: obsidian exciting nursing home of course, if one of the brackets does win, the ensuing litigation after the winner refuses to share and/or quicken refuses to pay out will be lulzy (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225214) |
|
Date: March 20th, 2014 9:28 AM Author: drab forum Who would the winner need to share w/? (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225219) |
|
Date: March 20th, 2014 9:35 AM Author: obsidian exciting nursing home if you sign up for a bracket using that website, the agreement is that if any one bracket wins, you share the winnings with everyone. it's a lottery pool.
this part would make it interesting too:
“If you get to the Final Four with a perfect bracket, I may buy you out of your position,” Buffett said. “I’ll make you an offer you can’t refuse. I will accept your phone call, you better believe it.”
someone signs up on that site and does reach the final four with a perfect bracket, then refuses Buffet's buy out offer. then their bracket fails. chances one of the other entrants sues? |
|
Date: March 20th, 2014 9:49 AM Author: drab forum you'd be an idiot to refuse his offer (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225286) |
|
Date: March 20th, 2014 9:53 AM Author: obsidian exciting nursing home (future plaintiff) (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225306) |
|
Date: March 20th, 2014 11:23 PM Author: diverse marketing idea jewess i read that his offer would be 50 mil plus tickets and hotel/transportation to the 3 final four games. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25230471) |
|
Date: March 20th, 2014 9:21 AM Author: rambunctious indian lodge gunner I was actually thinking that this would be a cool CS competition to put together (DE Shaw, or Google, or MIT)
let people write algorithms that generate a million brackets, if anyone can generate a perfect one, they win 10 grand or something |
|
Date: March 20th, 2014 10:32 AM Author: Vibrant Brethren They planned for the algorithm stuff. That's why they limited the number of entrants to 15M. Even if all 15M brackets were unique (and were the 15M most likely permutations, a priori), the odds of a payout would still be somewhere in the 1/1000 range. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225523) |
|
Date: March 20th, 2014 10:34 AM Author: rambunctious indian lodge gunner sure, I just think its cool to wonder if it could be done, and how few brackets it could be done in algorithmically (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225539) |
|
Date: March 20th, 2014 10:43 AM Author: Vibrant Brethren A supercomputer could generate every possible bracket in a little over a second. Give the teams preset strengths and it could then rank the brackets by a priori probability.
Side note: Not sure what all this chatter about needing to put it the number of upsets to be in line with historical averages, etc. is. The single most likely combination is the one with all favorites. Of course, that goes outside the realm of what you need to do to ensure that you have a unique bracket if you're trying to cop dat billion for yourself. |
|
Date: March 20th, 2014 10:58 AM Author: rambunctious indian lodge gunner chatter - so am I just being retard mathwise about that? (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25225668) |
|
Date: March 20th, 2014 11:34 PM Author: Cocky erotic market immigrant noob here. how many unique possible brackets are there? with enough collusion this sounds possible. (http://www.autoadmit.com/thread.php?thread_id=2520918&forum_id=2#25230545) |