How should Manifold encourage trading on long-term markets?

Basic

Ṁ1085

resolved Jun 5

49%48%

Loans of a fixed size per market (like before)

26%26%

Leverage: E.g. say when you bet $5, you pay $1 up front and $4 later (the $4 is loaned to you)

19%19%

Funds in long term markets grow (pay interest)

3%3%

Futures: derivative market Y on the probability of market X at time T

2%2%

Continuous partial resolution of "how many days until X" markets

1.7%Other

0.2%

Leaderboards for long-term markets

0.2%

Dedicated pool of money for long term markets

0.1%

Small bets have high margin: 100% of your capital, no leverage 25% of your capital, 2x leverage 5% of your capital, 4x leverage 1%, 8x leverage

0.0%

Pay interest on invested mana

0.0%

Insurance markets - "will X happen today?"

0.0%

Discourage them instead

0.1%

Demurrage (charge interest on uninvested cash)

0.3%

Market creators can offer dividends/bounties to draw in bettors

0.3%

Curated Long Term Tournaments with their own mana pool

Currently, trading on markets that resolve in a few days is far more incentivized compared to trading on markets that resolve in months or years, because you can earn similar amounts of profits much sooner. The M$ you put into a 1-year market could be traded hundreds of times on shorter-term markets for much more profits. This is reflected in the fact that we generally see more trading on short-term markets than long-term ones - although there are certainly exceptions to the general rule. Making predictions on long-term questions is very useful though! How can Manifold mitigate the disparity in incentives and help encourage more trading on long-term markets? This market resolves the market consensus (but I will adjust the probabilities according to my judgement if it seems to become too much of a Keynesian beauty contest). See related discussion on: https://manifold.markets/JamesGrugett/how-will-loans-be-implemented-on-ju Please do add more suggestions! May 31, 7:31am: "but I will adjust the probabilities according to my judgement if it seems to become too much of a Keynesian beauty contest" -> as pointed out by Matt, "Keynesian beauty contest" carries different meanings/connotations. So please instead read this as: if the trading on the market looks too much like pure price manipulation instead of actually trying to predict the question I asked, then I will adjust the resolution to ignore the seemingly-manipulated prices, according to my judgement.

This question is managed and resolved by Manifold.

Get

1,000

and

3.00

25 Comments

22 Holders

64 Trades

Sort by:

Resolved to market probabilities. I think Manifold automatically rounds them to the nearest %, so I excluded ones at ~0% - although I think those are good ideas as well!

Btw I didn't bet with an intent to make a profit, only to express my personal beliefs.

I don't think it's clear that all long term markets are not working well. Looking at the first two pages of markets sorted by 24-hour volume, so we can see the ones with most activity, I see 5 for end of this year, and 3 for 2024, out of about 20. That's actually more than I expected. I think it's ok if I only rarely come back to my long term bets. E.g. I like to make bets at the beginning of the year and revisit them at the start of the next year. And if I'm making 1 bet a week on a market that resolves a year later, that's the same amount of engagement as making 1 bet a week on a market that resolves next week, isn't it? (I actually am not a fan of the prediction platform overly incentivizing constantly updating my predictions, because it becomes a drag and I don't have time to keep them all up-to-date - this is a potential problem with there being too much activity on the markets!) And I do think that 1) having curated groups of questions is a good idea (relates to the communities idea others have proposed elsewhere) and 2) having the option to have separate mana pools for some groups of questions could be useful (I proposed this in another answer too). I think we can do both this and something that helps support those interested in long-term market creation and trading in general - they aren't mutually exclusive.

@jack Agree that having to constantly update can be a pain - it would still be good to receive regular updates/news on those markets, but that's an independant problem. Also agreed that this proposal isn't mutually exclusive. Selling my bet.

Tying up ones funds for a long term represents a big opportunity cost. Loans was tried as a feature, but I think correctly removed for making betting more complicated. Further, having lots of uncurated long-term markets which are abandoned and with very little activity after the first few days represents a less than ideal outcome for Manifold. Instead, curating long term tournaments (eg. US elections), where each tournament has it's own pool - subsidized by Manifold itself, or by some creator organization who will take it upon themselves to update the market with regular news and information - will solve both the tying up money and engagement problems.

@akhil I like this idea, but I also think Manifold's big differentiator is anyone being able to create markets. And if someone wants to create long-term markets, there are some inherent difficulties, but the more Manifold can help support those the better. Abandoned markets where the creator disappears we can deal with by https://manifold.markets/Austin/will-manifold-set-abandoned-questio or perhaps supporting some form of delegated resolution. Markets with very little activity are natural, but I don't think that's a problem - they will naturally show up only when they match someone's search, or there's been some new activity.

@jack Fair point. My thought process on this is that a differentiator is measured by how useful it is; and anybody being able to create a short-term market captures almost all of that value. The premise of this market, which I agree with, is that long term markets are not working well today. Loans have already been tried, and showed mixed success. If one wants to incentivize long term markets, coming up with the most appropriate mechanism that solves the probelm seems essential - even if it means not using the existing differentiator. Further, I think long-term tournaments curators could accept proposed questions from anyone, which at least partially takes care of your concern.

@jack Also, abandoned markets is I think less of a concern than markets that initially see some activity, but then have little to no engagement thereafter. If I'm making 100s of bets, my long term bets are quickly forgetten without a few committed users who regularly update that question with new information. Formalizing that role through curators for that market helps with this.

@jack Very much hope these are not brought back! Not worth the downsides.

@SneakySly What do you think are the downsides? I think the adjustments I proposed work well to deal with the main downsides Austin brought up.

@jack Subjectively they were unfun and "fiddly". Your proposed change seems even more fiddly. It would still be game optimal to use loans, but now I need to also think about the 10% requirement and how that factors in.

@SneakySly Yeah that's fair, I do agree on the fiddliness. I was kind of thinking of combining it with the up/down bet arrows (which are also kind of controversial, but seem overall to be popular) - basically your first click up or down (or maybe your first two clicks) can be paid later via loans. When you say it's optimal to use loans, what do you mean? Of course if you were already going to trade on a market it's financially optimal to do it with a loan, since it's interest-free. But that doesn't necessarily mean you should bet on every market - if you're betting optimally you still have to believe you have an edge over the market.

@jack There are markets where the answer has fully converged onto the correct answer already and there was not real money to be made. Without loans I can ignore them. With loans I should still put my loan money in and collect some fraction of a M$.

@SneakySly That's a good example, it does seem like ideally Manifold should (somehow) adjust the market structure so that those converge to exactly the correct answer with finite funds. Maybe the liquidity curve shouldn't be so extreme at the edges? If the correct answer is 50% (coin flip) it's cheap to move the price there, but if the correct answer is 100% it's infinitely expensive, and while there are some good reasons for it, the difference could probably be made less extreme?

A market creator can choose to subsidize a prediction market with a dividend, instead of or as well as liquidity. The dividend is paid out on a continuous basis to all bettors with a stake in the market, based on the mana value of their stake. Bettors compete to earn the dividend, but if they bet the wrong way then market movements wipe them out. Claim: dividends incentivize accuracy better than liquidity in long-term markets. Failure mode: someone makes two accounts and takes opposite positions on the same market to get a risk free return. Probably have to restrict dividends to verified accounts? Failure mode: the best bettors can still earn more on short-term market, so it attracts more second-tier bettors. I suspect "activist investment" addresses this some - a great bettor can move the market halfway to the correct probability and then add a comment explaining what the correct probability is.

@MartinRandall Hmm, interesting idea. I can see some ways it can produce better accuracy than liquidity - e.g. thinking about markets where we know it will resolve yes in 1 year, if we assume the traders have equal discount rates (not actually an accurate assumption), then currently the market will settle at something like say 90%, and the liquidity doesn't affect this - it only affects how much investment it takes to get there. Whereas a dividend does incentivize betting much closer to 100%. But it feels like there's some asymmetric/misaligned incentive here. If the market price is 50 and I think it should be 40, I can buy NO shares or sell YES shares if I already have some. I think those should be basically the same or else something is wrong with the market mechanism - but with this idea if I already have a YES position and sell it, I will get less dividends, whereas if I buy a new NO position I get more dividends. (I think this asymmetry is probably the same reason there's 's an arbitrage where you can take opposite positions on two accounts to get a risk free return.) Not sure if we can come up with a fix for it.

@jack Buying and selling is already asymmetric. Take your example of a long-term market that is at 50% but should be at 40%, but without dividends. If I have YES shares I should definitely sell them down to 40%. If I don't, then maybe I should but NO shares, but in a long-term market maybe I have better alternatives. If there are multiple markets with dividends then in your example it's probably still worth selling YES and using the mana to buy shares in another market with dividends. It does feel hacky, though, I'm not claiming elegance. Also, it suggests a model where long-term markets are restricted to creators willing to burn mana on dividends, which... maybe is okay?

@MartinRandall Oh yeah you're right that the asymmetry already exists. I guess this is also similar to the "long term markets pay interest/dividends" idea except that the payments are bounties provided by users. I think it's a neat idea to allow people to add higher incentives to markets they care about. Although if I wanted to subsidize a market I would not be sure how to decide on liquidity vs dividends. I don't think you'd need to require creators to put in mana for dividends, but I do think Manifold should do a better job of prioritizing showing markets with more subsidies/liquidity to bettors.

@jack one difference is that when paying interest I think we've been thinking of a fixed 2% interest rate. This idea has a dividend pool. When there are only a few bettors the rate will be higher. When there are many it well be lower. This has some nice properties.

@jack how about this: 1. Mana can be covered back into USD. 2. Someone has a long term question and knowing the answer is worth $1,000 to them. 3. They subsidize a prediction market with a dividend payment of m25,000, to be paid out to bettors in proportion to their positions, on a daily or continuous basis. 4. Bettors compete to earn the dividend, but if they bet the wrong way then market movements wipe them out.

This is weird and different enough that I'll add it as a separate response.

@jack I'll put some more mana where my mouth is. This is the simplest and most effective solution IMO (even though leverage isn't a bad idea overall).

I think this option is overvalued because it provides so little extra incentive that it'll barely register (as described in my earlier comment), and meanwhile it adds significant extra complexity and requires money from the platform that could be invested elsewhere. (I wish free response markets were CFMM so I could bet down.) Maybe I could be convinced otherwise though!

@jack yeah, way overvalued IMO.

@jack I like this idea by far the most!

Related questions

Related questions