Hacker News Reader

Comment by everettForth | Hacker News Reader

The expected value of an entry is something like ten cents assuming you know a lot about basketball.

This is like saying, "let's all work together to hack ten cents into eleven cents" (except we're doing something secret to do the hacking, which we won't tell you, but trust us. ps we love Nate Silver)

Edit: Actually, Now that I think about it more, the people who made this are probably concerned with too many teams using the same statistical techniques, and choosing the same bracket. Assuming several teams find this "highly likely" bracket, and it wins, those teams would all share the prize. This way, the organizers can choose their favorite bracket, and help ensure that other brackets aren't going to share the prize with them.

View on HN

I think your edit is where the real story lies. Assuming a given probability of Seed_i beating Seed_j, you could make sure that multiple brackets follow that distribution, rather than just picking the likely winner. By working in concert, rather than alone, you can cover the widest possible population of brackets.

The "highly likely" bracket is still only going to have astronomically small odds.

Yeah, "highly likely" is all fun and games until a college like George Mason comes in a destroys every single bracket.

Yea, I don't think there are too many "highly likely" brackets anyway--the first couple rounds are brutal. There are definitely "highly likely" Final Four's, but just one very minor upset on day 1 will ruin a perfect brack.

If you figure that the selection committee knows something about basketball and at least to a first approximation tries to slot teams correctly, it's relevant that the five seed has only beat the 12 seed 2/3s of the time and it gets worse from there. So you are already down to less than 20% odds of guessing the five seeds in the first round.

Or, it's a bit like giving away free lottery numbers, on the condition that a winner pays a very reasonable 10% of their winnings.

A gullible player will see it as great deal, with no upfront cost and no downside. A less gullible player will ask what contribution (none) the "seller" made to earn 10% of a win.

Well, in the scenario you describe, they're paying for the lottery ticket... they've clearly "earned" 100% of the wins.

He said lottery numbers, not tickets.

Ah, my mistake. In the number-only scenario, the value that person provides is organizational / financial. Every so often (though rarely), the expected value of a lottery ticket is greater than the cost of the ticket. Imagine a simple 50 choose 6 lottery, where a ticket costs $1, you choose six numbers between 1 and 50, and if you get all six right you win the jackpot. If, through the vagaries of chance, the jackpot reaches $30 million, you could buy one of each possible ticket (there are 15,890,700) and guarantee that you'll hit the jackpot. If, as is normal, the jackpot is shared among all winning tickets, then you have a decent chance of winning $14 million and an also-decent chance of losing $1 million or more, so it's still kind of risky in that scenario. But we can imagine the jackpot reaching, say $50 million, where even if another person gets a winning ticket you still make a hefty profit.

That's great if you have sixteen million dollars to hand, but if you don't, you could join a "lottery mining pool" ;) where you buy a certain number of tickets and the pool together covers the ticket space. In such a case, the person giving you the numbers is performing the valuable function of preventing collisions within the pool.

Fair enough, and a good point.

that's kinda paranoid thinking given the EV of the "best" bracket is probably like $1 instead of 10 cents. A lot of work to make a dollar.

It's a pretty interesting idea haha. if only there was some way (without spending a lot of legal money) to ensure entrants into the pool all split the money.

Reply on news.ycombinator.com