Do you think every market should also have a Metaculus-style aggregated forecast built in?
22
70
Never closes
Strong Yes
Yes
Maybe
No
Strong No
See Results

A companion to /Joshua/do-you-think-every-market-should-al

Sometimes a market is not the best way to get the most accurate prediction for something. In the 2023 prediction contest, the Metaculus forecasts outperformed not only Manifold but also almost everyone else.

I think there are lots of reasons not to take this at face value, the biggest one being that Manifold was just a lot smaller at the start of 2023 and I think we've gotten more accurate since due to both growth and other improvements to the platform.

But nonetheless, this is striking! And it makes sense for a number of structural reasons.

My proposal is that obviously if a pure aggregated forecast can be more accurate than a market, then every market should just also have an aggregated forecast built in. Metaculus doesn't own the concept.

It could just be another tab on every question, that everyone can reference for trading.

Get Ṁ200 play money
Sort by:

I voted NO because I fear it will confuse people. Of course, it can be hidden by default but then it probably doesn't get much use. Is it useful enough just as a tool for a few power users?

I'd love to have an interface where you can put in your own probability estimate, then the interface gives you some options:

  • Bet to your percentage (and how much mana it costs)

  • Bet 10/50/250 mana in the direction of your percentage

  • Some kind of Kelly bet?

Meanwhile the percentage you entered gets aggregated metaculus style whether or not you also choose to bet on it

@Fion These are good ideas! I also think it would be very helpful if you could search through markets in which you've made an estimate by how much mana it would take to move the market to your probability. Then you could see where there's the most profit to be made if you're right.

@Fion In other words, integrate https://manifol.io/ into the site.

So the idea is that in addition to bets, people can give their actual underlying probabilities for the event occurring? And would the main advantage be that, if someone doesn't bet because the market odds are similar enough to their own, they can still give their odds and improve the aggregate forecast?

Maybe there's a way to integrate this with limit orders somehow, like where you give an "under" and an "over" probability where you want to buy or sell, and your actual probability is between those two

@CDBiddulph Yeah, I envision it as just a separate tab next to the market probability. You click that, and then it switches to a metaculus-style interface where you just put in the probability you think the market /should/ be at.

I'm all for telling people to bet on their beliefs, but we don't have to be absolutists about it. I think lots of markets should be at 99% or 1%, but I don't buy them to that percent because of interest rates.

There are also many situations in which markets are better that forecasts, but we don't actually have to choose! We could just have both! If I were an active metaculus user, I'd be lobbying for them to add in markets for all of their questions.

It just seems like it would obviously make both forecasts more accurate.

More related questions