![](/_next/image?url=https%3A%2F%2Fstorage.googleapis.com%2Fmantic-markets.appspot.com%2Fcontract-images%2Fgpt4%252Ff70d451b1f2c.jpg&w=3840&q=75)
Edit: The answer is No and a little complex. See the comment thread below for the details.
I'm creating this poll mostly to understand how Manifold work.
Edit: Simpler example:
I create a market, it has only 1 trader who bets 1000 mana on YES. I resolve it legitimately as a NO. Is that 1000 mana effectively deleted from circulation?
Original Example:
Say there are two scenarios:
Scenario 1: A market with many traders reach 50:50 odds and then resolves to YES.
Scenario 2: The same market, but just before it resolves, a large trader comes and shift the odds to (approx.) 100:0 YES:NO, netting a large earning.
If I understand how things work, the large trader has not affected any of the earning of the existing traders, so where is the mana coming from? Without reading into the math behind betting markets, I see two scenarios:
The mana comes from the pool traders have created. This means that Scenario 1 effectively deleted that mana.
The mana comes from Manifold subsidizing things which means that the amount of Mana generated in the market changes based on how the market closes.
If Mana is indeed deleted when closing any market, did anyone estimate how much is removed every day (vs created)?
Related questions
@Stralor Notably this happens silently, so creators have no idea they get a lil refund if they aren't watching their balance like a hawk
@Stralor Thanks!
How does the subsidiary reclaim work? Would it be 100% of the subsidiary if the market is at a hypothetical 0:100 and resolves to 100:0?
Does this also mean that the hypothetical last investor has only a portion of the subsidiary to earn, and that the amount of mana invested so far in the market doesn’t matter?
@gpt4 Now we're digging into what the math of it all does. Manifold has an "AMM" ("Automated Market Maker"?), which acts as a counterparty to all trades that aren't against other users' limit orders. The AMM gives odds based on current subsidy, amount bet, and market value (%).
This is important for your second question, because yes, the last investor has only the remaining unclaimed subsidy (& limit orders) available for the taking. However, the amount of mana invested does matter bc the "House" (Manifold) adds more subsidy to each question based on how many traders it has built up, and it does so automatically. NB: added subsidies, whether from the House or not, trickle into markets over time so a trader can't show up and immediately capture their own value-add for profit inflation.
Go to a Yes/No question and hit the context button on the top right (the three dots). It'll show a lot of stats about the question, including "Pool" which iirc is how much mana is directly available for capture from the current position to [100%, 0%].
![](https://firebasestorage.googleapis.com/v0/b/mantic-markets.appspot.com/o/user-images%2Fdefault%2FYixyu7C3jD.png?alt=media&token=064065e6-6684-4048-8410-d33a4b60166d)
To the first question: yes, a complete inversion of resolution would quietly return the subsidy to all subsidizing parties. Of course, in practice that's very rare except for misresolves which are often reversed and punished.
@Stralor I’ve tried looking at this market (which is at 99%) https://manifold.markets/LandynMcAfee/will-bitcoin-reach-60000-in-2024-bf0412b0af0b?r=Z3B0NA
The subsidiary is listed at 450 mana, but if I bet a hypothetical large amount of mana on a NO, I can see that I can capture all of the liquidity on the market (almost 20k mana). Where is that coming from?
@gpt4 I believe it's coming from all the share purchases made by YES traders, which at this point are often quite inefficient. At 99% they're paying 100 mana for 100 + 1 shares, but the "other" 99 mana doesn't just vanish without generating countershares. I'm not 100% sure on that process however.
@Stralor So my original question still stands then. If that market resolved as a NO tomorrow (assuming it was a legit NO), would all that mana disappear? If that’s the case, then mana is indeed removed from circulation when you close any market (outside of a hypothetical 100:0 perfect market).
@gpt4 in short: no, all of the mana provided by users (not shares) is recouped by some user. in a YES/NO outcome, shares only equal mana output when their outcome is picked, otherwise they're worth 0 mana. the rough idea: at 50% 1 mana in creates 2 shares, split over YES and NO by market value, of which the buyer gets one type and the other type is virtual, available to other traders.
to get more in-depth (though maybe more confusing), let's look at a case with fractional shares! sometimes a market will resolve to something other than 0 or 100%. If it resolves to, say, 20%, then each YES share will convert out to 0.2 mana and each NO share will likewise pay out 0.8 mana. And of the remaining 20% of the subsidy, the creator will recoup a fraction based on how much of the subsidy they provided during creation vs. all other subsidizers (the House & users who opted to directly add subsidy with their own funds), who will also recoup the rest equal to their contribution's share of the total. -> if I put in M1000 subsidy on a market that ends at M2000 total, I can expect back half of what's remaining unclaimed.
this is probably a bad time to mention it, but multiple choice markets are even more complicated and do currently have some subsidy loss due to bugs.
@Stralor thank you for your patience in explaining, it actually helps a lot in understanding how the prediction market works behind the scene
However, I still don't think that it contradicts my hypothesis that mana is destroyed whenever you close a market (and more mana is destroyed the further away the percentage was from the resolution percentage)..
Here is a much simpler example to help explain the problem - I open a market. You bet X mana on YES. I resolve it as a NO. There are no other traders in the market. Where does your X mana go?
@Stralor Just tested it. You are correct. It transferred all the betting mana + all the subsidy to the question creator. This means that creators are heavily incentivized to stop trading on the question before the event actually happens.
In my opinion Manifold should just remove that mana from the marketplace. Creators are already incentivized to create questions by getting mana for each trader, and I doubt that many creators knows of this effect. This is also a slightly negative effect as it biases market creators (given that the way they resolve the market affects their payouts)
In addition, for Manifold to actually make money, they'll need actual mana sinks, and this seems like a very straightforward one that will make every market give back a bit.
@gpt4 Interesting perspective. I don't 100% agree, and definitely as a creator I see benefits in engagement, bonuses, and reputation from leaving my markets open, but it's a valid point!
@gpt4 I think my biggest argument against this is: when does it happen that a creator who knows about this will use it to their advantage? By far most instances would be misresolves and against the culture and guidelines of this place. If creators want to design markets that close early, go for it! But it's an uphill battle to win over traders when there's a duplicate they can live trade on.
@Stralor it just makes the whole "I won't bet on my market so I won't have a conflict of interest" a flawed premise - the creator still has an incentive to resolve it against the dominant direction (or at the very least, to make things more ambiguous, that that the market remains further away from 100% to one direction.
@gpt4 true! but I believe other incentives are stronger, like reputation, league position, profit chart, etc.