How will "range markets" work in Manifold?

Basic

Ṁ339

resolved May 12

ResolvedN/A

9%Other

45%

Using CFMM, allow market creators to map 0-100% to a custom scale of their choice

Using DPM, allow market creators to set up buckets (e.g. options of 10-20, 20-50, 50-200)

1.1%

(ante & discussion thread)

31%

Using CFMM, allow market creators to set up buckets

Using LS-LMSR (aka the Hanson-Othman mechanism)

0.9%

CMMM - Constant Mean Market Maker: Uniswap CPMM generalised to more outcomes

0.4%

Moving Normal Distribution

A popular request is for Manifold to natively support "range markets" aka scalar markets: markets where you predict from a continuous range of numbers, rather than a binary YES/NO or a percent from 0-100%. See also: https://manifold.markets/Cyril/will-manifold-implement-range-marke This market will resolve to the option that is implemented as of May 1st, 2022; if none are implemented, than the one we devs have decided to use at that time; if we haven't decided, this market resolves N/A.

This question is managed and resolved by Manifold.

Get

1,000

and

3.00

11 Comments

8 Holders

22 Trades

Sort by:

No objections here.

Hm, at this point in time I think "Using CFMM, allow market creators to set up buckets" is our leading proposal, but we don't have native support for it. The thing that is the current best fit is "Using DPM, allow market creators to set up buckets (e.g. options of 10-20, 20-50, 50-200)". Some people still do "Using CFMM, allow market creators to map 0-100% to a custom scale of their choice" but I'm a bit suspicious of the mechanism I'm leaning to resolve to N/A, any objections?

Probably converting existing 0-100 questions is only possible if this is done, and my question on whether I'll be able to convert them is at 16% https://manifold.markets/Tetraspace/will-i-be-able-to-turn-my-existing-18f22bbe8829

I still don't really trust the % scale formulations. I don't think i will till it's trivial in the UI to *figure out my profit conditional on arbitrary Prob resolutions*, esp. at purchase time.

Seems simplest as I would imagine it’s mostly presentation logic beyond the setting of the 0% and 100% values (with of course some math in between).

This accurately summarizes my view as well. I think it allows for easier dissemination and comprehension by the general public as well as it most closely matches visually how a stock market is presented.

This is a crackpot idea. Half baked. So you have a range say number of pushups I will do this morning. Pick a mean say in my case 3 and a probability of doing double, say the 50%. A normal is created based on this. Now when you trade you buy a range of your choice. Ill have 3-4 pushups and it works out the probability. Youd need to shift the mean and variance somehow along with the size of the bet to create that slippage we all love.

Mentioned in https://medium.com/bollinger-investment-group/constant-function-market-makers-defis-zero-to-one-innovation-968f77022159 I doubt you would go for this as it would feel different from C(f-Pepe)MM used for yes/no. But I see this market as an ideation session.

FWIW, I lean towards "betting higher/lower than current amount" being a much better user experience compared to "decide which bucket your guess falls into". Buckets also put a lot of burden on market creators to find sensibly-sized buckets.

"+ initialprobofC * ln(Csharesontable)", and so on.

Just as binary CFMM keeps constant "initialprobofyes * ln(yessharesonthetable) + initialprobofno * ln(nosharesonthetable", a known-in-advance-multiple-choice CFMM can keep constant "initialprobofA * ln(Asharesontable) + initialprobofB * ln(Bsharesontable) + initialprobofB * ln(Bsharesontable)".

Related questions

Related questions