Resolves to YES if CNN declares that Harris won the 2024 Presidential Election, and 24 hours later the price to buy Harris on Polymarket before fees is at least 97.0.
Resolves to NO if CNN declares that Harris won the 2024 Presidential Election, and 24 hours later the price to buy Harris on Polymarket before fees is below 97.0.
Resolves to Fair Market Price if CNN instead declares anyone other than Harris to have won the 2024 presidential election.
Only the first declaration counts, even if they take it back or reverse it.
(To be clear, if the market outright resolves during the 24 hour period, I will consider that a price of 100 or 0 cents accordingly.)
1,000
3.00@ZviMowshowitz Is there a writeup of the advantages of this system over N/A somewhere? I'm sore because I didn't make a wrong prediction yet I lost money :(
@PaulCrowley I'm sympathetic to Zvi's approach on this. Without access to the relevant counterfactual (Kamala winning) the market price is a fair estimate for how likely a YES in that counterfactual world would've been. And we do have evidence to update on since this market was created. Like seeing that polymarket hewed to CNN's declaration (I presume). So "mostly YES" feels like how reality played out.
That said, "resolve N/A if condition not met" would be less confusing. Still, all in all, I think I like this better. Slightly related: my anti-N/A screed.
@dreev yes, but it also punishes you for betting against consensus, which is strange (the whole point of prediction markets is to reward that!). the market price reflects the consensus of participants weighted by capital. i have 1.5M mana. if i were sufficiently interested in this market, the opinion of paul or you or zvi would be irrelevant, it would reflect whatever i bet my mana on (until a jackson or etc comes along, and then my opinion would in turn be irrelevant).
which is fine, because if the "consensus" price i pick is bad, it is profitable for you to bet against me. sweet. but if you resolve to market price rather than N/A, that means that your bets can become unprofitable based on future market activity, even if your bets were well calibrated/correct compared to the market price (especially if the non-N/A case is very rare, it's possible for the +EV you gain from a bet that's more accurate than the market price to be erased by the market movement).
i have always found that system genuinely bizarre. with N/A, you can place bets and precisely estimate whether they are profitable (accounting for interest rates) purely based on your own forecast. with "resolve to market price", to estimate if a bet is profitable you instead need to predict the future beliefs & betting activity of whales, a rather strange (and irrelevant) exercise.
@Ziddletwix Ah, this is the debate about self-resolving markets. Zvi stipulates in general that he'll only resolve-to-PROB if he's satisfied that the kind of shenanigans you're describing are not in play. Of course that's tricky to formalize but seems to work ok in practice. @jack has more thoughts on this: https://manifold.markets/post/selfresolving-markets-why-they-dont
(I'm still interested in the question of more general ways to make self-resolving markets work.)
@Ziddletwix Confirmed what Daniel says, if you tried that I'd simply resolve to a different number or if necessary N/A the market. So far I've never seen an issue, because I make it clear in my house rules that I'll do that if I have to.
Zvi stipulates in general that he'll only resolve-to-PROB if he's satisfied that the kind of shenanigans you're describing are not in play.
yes, if I buy it to 99% and hold it there, of course zvi can easily exclude that non-legitimate activity. i'm not referring to that. i'm referring to the natural market price (i.e. what people actually believe, not attempts at manipulation), reflecting the consensus of participants weighted by capital. i'm only using the example of "i personally have lots of mana" because it's easy to think through, but the concerns are equivalent if that market price reflects jack & i having a bunch of mana or any number of bettors who contribute to the final equilibrium.
prediction markets are designed to reward people who (accurately) bet against consensus. resolves-to-% has the (imo unfortunate) property that if you are a sharp forecaster and place a bet that is +EV at the current price, that bet might not be +EV if it turns out the consensus of other future participants sufficiently disagrees (especially if there's a high chance of no resolution).
e.g. for this market, under conditional-N/A, to calculate whether a bet is worthwhile all I need to know is P(YES | Kamala), P(Kamala), and my personal interest rate. under resolves-to-%, my forecasts of P(Kamala), & P(YES | Kamala) could both be correct (& superior to the consensus) and yet the EV of that bet hinges on the activity of future traders.
markets on manifold are not very deep, & are frequently inaccurate (maybe the extra context here is it becomes harder to have much faith in the market price when you have lots of mana lying around & you start to push it around). i much prefer giving people the tools to calculate whether bets are worthwhile based purely on their own forecasts, not punishing them if they diverge from consensus (without them actually being wrong).
@ZviMowshowitz it's not just about people doing obvious manipulation. Ziddle makes a great point that in the roughly half of worlds where the conditional isn't satisfied, you are paid for predicting what other people think.
For example, if I think that the conditional is 50% likely, and my prediction if the conditional is satisfied is only 60%, but I predict that other people think it's 95%, then it's +EV to buy YES if it's 70%. So the market fails to capture my actual beliefs. None of this ends up triggering any rules about the market price being unfair.
Polymarket rules state that they will resolve based on media projections.
The resolution source for this market is the Associated Press, Fox News, and NBC. This market will resolve once all three sources call the race for the same candidate. If all three sources haven’t called the race for the same candidate by the inauguration date (January 20, 2025) this market will resolve based on who is inaugurated.
The election denialism last cycle was a lot more concerning because Trump was commander-in-chief at the time and maybe he could have actually pulled off an autogolpe.
This time that's not the case and we know the supreme court didn't put up with his nonsense last time. There is also a lot more smart money on prediction markets this cycle.
From his house rules:
Ambiguous markets that have no other way to resolve, because the outcome is not known or situation is truly screwed up, will by default resolve to the manipulation-excluded market price, if I judge that to be a reasonable assessment of the probability involved. This includes conditional questions like ‘Would X be a good use of time?’ when X never happens and the answer seems uncertain.
So Zvi will attempt to exclude "manipulation", and this should not become a whalebait (at least in the "last minute large trade" sense).
(FWIW I disagree with this approach & I think the N/A resolution is a much more clean solution, but at a minimum I wouldn't worry about someone buying this to 1% to suck up the value in the case that Trump wins, Zvi would exclude that.)