The Keynesian Beauty Contest is a game theory experiment where a group of players are asked to guess a number between 0 and 100.. The winner of the contest is the player(s) who guess closest to half the average of all the guesses.
In this version, the options increase exponentially up to more than a million. I believe this will make the market significantly more dynamic, and greatly increase the incentive to "fight" over the final resolution.
I've also made the market only 2 weeks long, since so much mana ends up locked up.
This is an attempt to perform this experiment on Manifold, but using continuous averaging of options. At close I will resolve the market to a linear weighting of the options above and below half the average of the unrounded probability weighted value of the options OVER TIME.
For example, if all options are bid down to 0% except 32 and 64, and 32 stays at 30% the whole time, and 64 stays at 70%, the final result is (32*.3 + 64*.7) / 2 = 54.4 / 2 = 27.2, which would mean this market would resolve to 30% 16, 70% 32.
Obscenely unnecessary? Wildly overcomplicated? We're all nerds here, deal with it.
UPDATE
First version of the code: https://pastebin.com/R8FXseEG
Go search for bugs: /DanMan314/will-someone-find-a-bug-in-the-firs
This may not be the final version I use to resolve the market. I'll re-run it when I feel like it, and post the output - I'm not holding myself to any particular schedule.
Changelog:
Updated to account for linear weighting of non-adjacent options
Output 12/14 10:50AM PST
Total Bets: 27323
Current Weighted Sum: 33146.69254685717
Time Weighted Sum: 31718.00963213971
Final Answer: 15859.004816069855
Resolution: 6% 8192, 94% 16384
1,000
3.00@DanMan314 your code has been matching pretty closely to an independent check that I've been doing with a spreadsheet. If there is a bug, it probably wouldn't affect the answer very much.
On December 9th, my balance suddenly and inexplicably dropped by about Ṁ2000 (i.e. from Ṁ600 to -Ṁ1400 or thereabouts). Unranked profit dropped by a similar amount, but ranked profit did not drop. To my knowledge, no questions were resolved (or changed resolution) at this time. Did anyone else see something similar or have any idea why this might have happened?
@DanielParker There was a bug in loans that the admins are now clawing back:
https://discord.com/channels/915138780216823849/994321335419420832/1182876492057825320
made some edits to the posted script. Seeing this as the expected payouts given the current probs and assuming they're static:
Resolution: 93% 16384, 7% 32768
Resolution: 70% 16384, 30% 32768
Edit, had a bug that discounted the forward looking weights too aggressively:
Not sure if that implies there should be additional upward drift as the 8K line is bet NO.
As @BoltonBailey points out below, the 4K line is already impossible to profit from. The 8K line will soon be impossible to profit unless someone takes a huge position in 0/1/2/4/8 lines to drive the average down at a loss. The 8K line is still part of the resolution barring >2% avg on the 1M line till the close.
@TrentonPotter Your investment in 32K hangs on not enough people wanting a risk free return of 1% in a few days
@Simon74fe Yep. More precisely, it's that the risk-adj return on the 32k line will have more leveraged capital behind it than the 1M line NOs. (@1%, you need 1/50th the capital for 1M and 1/50th for 32K with a yield of 2-5x on the 32K bet).
It's just super capital intensive to get behind the NOs, even against new players w/ small banks. The risk of someone with 'real' money moving into the same position is a risk on the 16K line
@TrentonPotter That's true, but if you buy YES you're burning Mana. At some point it will no longer be worth it to burn so much Mana
@Simon74fe Yeah, just flipped my position. I don't have a large enough bank to compete w/ u and Fion. (The YES you burn on the 1M line is made up w/ the 32K, but you guys have >100x networth and there's more traders entering the market).
The illiquidity assumption I had was WRONG.
@Fion To put it a different way: I am willing to pay hundreds of mana (i.e. I've spent ~Ṁ440 thus far) by betting YES on the higher numbers in order to get the final number to resolve closer to 16,384 (which is where my main payout is where I stand to win ~Ṁ1802). There's a limit on what I'd be willing to pay to artificially keep the number up, but for now it's still profitable.
By my calculations, it's already impossible for 32 (or lower) to have any resolution share
@BoltonBailey By my calculations, it's now almost certainly impossible for anything lower than 512 (or higher than 524288) to make a profit (i.e. assuming a 50/50 vote split between 0 and the expected resolved number(s)).
@DanMan314 The comment is about 5096 not being a power of two - and consequently none of the numbers higher than that are either.
(It's really distracting - you think you see your old friends, the powers of two, but instead you see some weird mutations of them.)