I've attempted to include approximately the top 35 most likely movies, judging from different folks' lists.
Will resolve equally to all films that are nominated for best picture. In other words, if all 10 films nominated for best picture are in this list, I will resolve each option to 10%.
I've been told that free-response will be added to multiple choice markets at some point, at which point I will allow new movies to be added.
1,000People are also trading
N/Aing this market - sorry for all participants.
The biggest issue with this market is that it's a multiple choice market, but it's resolving equally to all nominated movies. Thus, it means that if 10 movies were nominated from this list (which seems quite possible), it would resolve equally to each market at 10%.
What this means in practice is that the market values for things like Oppenheimer are way too high. It could only justify an 18% chance if you think only about 5 of the movies in this list are going to get nominated for best picture, which is likely way too low.
This is exacerbated because Manifold's UI wasn't intended for this use case, and so, when betting on Oppenheimer at 18%, users are told that they can win up to 5x their investments (which is wrong).
I might reconstruct this market with the new multiple choice formats, but my conclusion from this is that in general, multiple choice markets where the probability resolves equally to all participants are not a good fit for manifold now.
@DylanSlagh yeah I think I'll do that. This market was an experiment with the MC format for "multiple answer" questions, but I think it's clear it does not work out very well.
The title now seems to disagree with the market description.
If five films from the list are nominated and you haven't been able to add any new films to the list, do they resolve to 10% or to 20%?
That's a really important difference when a film that I think is very likely to be nominated is trading at around 12%!
@SimonGrayson I considered this, but I expect that free response will be added before the next Oscars, so I expect all options to be there. Also, the title is limited in characters.
@chilli Just want to clarify: if there are fewer than 10 nominees, are they going to resolve equally, or to 10%?
@chilli Is it guaranteed? I thought that was something just changed for last year, and they could well change it again.
https://en.wikipedia.org/wiki/Academy_Award_for_Best_Picture#Nomination_limit_increased
@jack generally rule changes persist until they’re removed. In this case, the rules for the 95th Oscar’s have already came out and I haven’t seen anything about reducing the limit. But yes, in the case that there’s less than 10, they will resolve evenly.
@chilli New markets support adding new options, but old markets do not (because the math doesn't work and it could generate unlimited mana).
So, what do you plan to do for this market?
Yeah, I think either N/A or resolve equally could make sense.
It could still be reasonable to go with the stated critiera:
Will resolve equally to all films that are nominated for best picture. In other words, if all 10 films nominated for best picture are in this list, I will resolve each option to 10%.
I've been told that free-response will be added to multiple choice markets at some point, at which point I will allow new movies to be added.
This is totally fine on its own, the only problem is the title.
@jack I think the bigger issue is that the "resolve to 10%" format has ended up very problematic for these multiple choice formats.
I'd really like a market format that automatically ensures that the sum of all probabilities sums to say, 1000%.
@chilli Yes, that has always been problematic on these types of markets, but you made it as clear as possible and on balance it's better to have these imperfect markets than to not make them at all.
They are likely coming soon! https://manifold.markets/jack/will-manifold-add-multiple-choice-w
@jack I'm not actually convinced that on balance it's better to have these imperfect markets. I think they confuse users of the site and directly causes mana to flow from those newer to the site to those more familiar with the details of Manifold markets.
@chilli Given that the last 5 trades as of now are people buying shares in markets over 10%, I would suggest that (whilst this is not absolutely necessary for those trades to be rational), other people have probably missed the bit about it resolving in a non-standard way. Maybe put it in the title? The problem I think is that 'has a film been nominated for best picture' is so clear and objective that there is less reason than usual for people to look at the resolution criteria.
FWIW, for future markets, instead of resolving equally to all the correct answers, it may be better to structure it like resolves each correct answer to 10%, and the remainder goes to a "Less than 10 correct answers" bucket - like this: https://manifold.markets/jack/what-markets-will-i-consider-improp
But also, Manifold plans to support multi choice markets that don't sum to 100% in the future, and that will take care of this better.
Killers of the Flower Moon looks underrated to me: 9/26= (just under 35%) of Scorsese's films have been nominated. And in fact the base rate is probably an underestimate here because a) that includes years when the number of best picture nominees were lower, b) Scorsese's career goes back to the early 70s, and yet 6s of his nominations have come since 2000 whilst only 9 of his films have come out since then (so 2/3rds since 2000 received nominations), and c) this kind of big, serious, moral, but relatively commercial, genre-y, epic seems like the kind of thing people would see as "best picture worthy".
@DavidMathers Good analysis! Note that the current prediction is actually roughly 100% for it to be nominated, because:
Will resolve equally to all films that are nominated for best picture. In other words, if all 10 films nominated for best picture are in this list, I will resolve each option to 10%.
@DavidMathers The payouts assume a 100% resolution, since that's more common. The only way betting above 10% can be profitable here is if there are fewer than 10 nominations.
@jack Wait then why did you just buy $5 of an Oppenheimer when it was already at 11% if that's a guaranteed loss?
@jack So is 'The only way betting above 10% can be profitable here is if there are fewer than 10 nominations' right or wrong?
Mira is almost right, except it's not the number of nominees, it's the number of nominees that are listed here.
@jack I'm also predicting you'll be able to add extra options by the end of June, so with respect to "Mira is almost right" - I don't recommend trading assuming the list will be static for 1 whole year.
Edit: Everything I've said here is confused and wrong, leaving it just so that the conversation makes sense.
Seems like buying NO on a market which resolves to multiple is a pretty bad deal, though (if it's not for selling YES shares). Like here for example if you think some movie is 30% likely to be nominated and its current probability is 10%, then you can't correct it by buying NO unless you think people will correct it further and you'll be able to sell - if it goes to a resolution, you'll only get the profits for a 10% NO bet. Maybe we just shouldn't make any such markets before we get the option of probabilities not summing to 100%.
@NamesAreHard If 10 get selected those NO shares will payout 0.9 I think. So in your example, NO would get 0.9 30% of the time and 1.0 70% of the time, EV 0.97, so you can bid it down to 3%.
@NamesAreHard Yeah, perhaps I should have waited. 🤔
Oh well, I'll make another market with the new format (or modify this one to have the new format) when they happen.
Edit: Everything I've said here is confused and wrong.
Huh, wouldn't have expected that. Where would all the additional mana to pay out come from? If I understand it correctly, the options here are kind of isolated from each other (with some arbitrage mechanism to make them sum to 100%) and you can think of them as separate binary markets.
So if option X is at 10%, then the bets are balanced at that percentage - you get 100 YES shares with only 10 mana and you need ~90 mana on NO for the same. If it's then supposed to payout NO as if you had bet at 90%, then the market will need to pay out further 90 mana in profit when there's only 10 available from the YES shares.
@NamesAreHard If an option resolves to 10%, then buying NO above 10% or buying YES below 10% are both profitable, and the incentives are for the market to buy it towards 10%.
The mana mostly comes from people betting against each other. If someone buys YES at 10% and someone else buys NO at 10%, they are effectively trading with each other, intermediated by the automated market maker. This is indeed basically the same as on a yes/no market. Note that a yes/no market can also resolve to 10%, and the market math there is exactly analogous - the incentives correctly make it profitable to bet towards 10%.
So if option X is at 10%, then the bets are balanced at that percentage - you get 100 YES shares with only 10 mana and you need ~90 mana on NO for the same. If it's then supposed to payout NO as if you had bet at 90%, then the market will need to pay out further 90 mana in profit when there's only 10 available from the YES shares.
In this example, the error is that you spend 90 mana to get 100 NO shares. And the NO shares payout 90 mana. So neither the YES nor the NO shares payout any profit, as expected because the trade happened at 10% which was the same as the resolution value. There is no "further 90 mana in profit".
If that answer doesn't end up getting chosen, i.e. it resolves to 0%, then the 100 NO shares payout 100 mana, so you get 10 mana profit. Which came from the YES trader who bet 10 mana.
@jack
Edit: Everything I've said here is confused and wrong.
Agreed about the example, I was trying to explain why it can't work the way Travis suggested (or maybe I misunderstood their suggestion because I'm very confused by the example calculation there but they did say NO pays out at 0.9).
Was this a reply only to my last comment? Because I agree this is how it works for an option that we know for certain will resolve to 10% (you're making profitable YES bets up to 10% and profitable NO bets down to 10%). But my initial comment's point was that the NO bets have different dynamics when the outcome is uncertain and even buying NO from 20% to 15% here in this market is unprofitable if you never sell (unless you think the actual probability of the movie being nominated is less than ~15%).
@NamesAreHard What they are saying is each no share pays out 0.9 - not profit, just payout. If you bought at 10% then you also paid 0.9 per NO share. Their comment
So in your example, NO would get 0.9 30% of the time and 1.0 70% of the time, EV 0.97, so you can bid it down to 3%.
is exactly right. For each NO share you bought at 10% for a price of 0.9 each. If it is nominated, it pays out at 10% i.e. 0.9 per share, and you don't profit but you also don't lose anything. If it is not nominated, it pays out 1.0 and you profit.
@jack Oh, I see! Thank you for the explanations, for some reason I had NO shares losing everything in the nominated case... Disregard everything I said here, market is fine :) That's probably the most stupid I've been on Manifold so far and I'm so glad it's in the comment section and not when betting lots of mana on a market :D