Will Manifold fix betting before the end of 2024?
Nov 2

During an exchange with @FubbiQuantz, I posted the following:

Intelligent bets are overrun by people dumping mana on positions in between when information is announced and market is resolved.

The standard for a prediction bet is a window where betting is allowed to increase fairness between early betters and late betters. Manifold ignores this standard because having more markets "open" makes the site look bigger than it is.

Manifold marks markets they don't like as non-predictive excluding them from leagues. However, they are happy to leave markets like this in the competition pool:


That market has a 10,000 M bet after the market should have resolved YES. This distorts the systems purpose and ruins competition for non-whales. The problem could have been avoided with a close date.

This market resolves YES if two things happen:

  • Prior to Dec 31, 2024, close date becomes mandatory on "Predictive" markets.

  • This is communicated in the UI for opening new questions.

This market resolves N/A immediately if Manifold rules it "non-predictive".

To prevent sniping, betting will close Nov 1, 2024.

Get Ṁ600 play money
Sort by:
predicts YES

I would like to thank @PC and several other people who have begun taking this more seriously. I apologize if the initial crudeness of the argument was off-putting.

For those whose reaction was to accuse me of bitterness and to conjecture no problem existed, I'm considering closing my current position at a small loss to avoid the potential of hypocrisy. However, the large bet was followed by the discussion (but not yet action) I wanted and I do not know if the discussion follows from the bet.

This is the worst one I've seen.


Unclear resolution criteria so nobody's sure whether it will resolve yes or no and then the market's creator comes along and bets 8000 just before resolving. The way he bet made it look like he was trying to fix calibration scores more than just making mana.

I don't think there's any simple way to create an open system like this that doesn't have loopholes for cheaters to exploit. If the people running the site actually want to make these fair markets then I think a set of fair play rules, with options for us to report bad players and penalties given by moderators. But somehow I doubt that's how it will go

predicts YES

@Nosaix Moderators are already manually checking markets and marking some of them as "non-predictive" preventing them from being scored in leagues. I learned of this when one of my markets was marked as such. For that market in question: I actually agree that it's "non-predictive". It was a literal shitpost over some of the frustrations discussed here.

My frustrations are as follows:

  1. Markets are being moderated. Moderation describes an appeal process that is approximately at-mention someone and hope they change their mind. That's not practical for scaling.

  2. Markets are not being moderated fairly. Your example if it does not contain factual inaccuracies is a good one. I don't know if it does or not but as I have pointed out questionable market behavior from @MarcusAbramovitch before I'm not inclined to waste more time confirming my viewpoint from someone's who bio contains "Going to be spending a lot less time on Manifold going forward.". <sarcasm>Your example is an entire week old.</sarcasm>

  3. Even if markets were perfectly moderated, mana gained in "non-predictive" markets is legal tender (and therefore leverage) in "predictive" markets.

this market seems related. A rollback tool could also help creators only honor predictive bets rather than after the fact trading. It should be low technical debt to implement, the question is whether it could be abused if not restricted to admin use.

The way to fix this is have markets where it takes say a day for bets to go on. If the info comes out all bets from before it became public get revoked.

@NathanpmYoung I personally think it would just be nice to see a breakdown of profit before and after info comes out. That being said, how do determine that time, esp with insider knowledge...

I've thought about this issue a few times, but I can't think of an easy way to resolve it. Closing markets way in advance leads to people's mana being locked up, and that tends to make people upset - but leaving them open leads to markets reaching 99.9% or 0.1% by the time they close. Being able to set randomized closing times might help discourage sniping, at least a bit.

I think the current system works okay for the most part, though. I've made a significant amount of profit while focusing primarily on predicting on long-term markets, and my calibration bears that out. If I want proof that I'm good at forecasting and not just sniping markets, I can always point to that and my Metaculus profile.

@evergreenemily yup as I mentioned in my comments, the annoying thing with closed market is the mana gets locked up

@PC This is me spitballing, so I'm genuinely asking, not stating my position. What would be the harm in letting people sell their positions in closed markets?

@JamesColiar Honestly, this would make the experience so much better. No need to ask creators / Manifold to resolve things, which is a pain for everyone.

The main drawback that I can think of is that there might be a bunch of unresolved markets

@PC To the point of @TravisLuckenbaugh and this market, I would love for there to be predictive profits and non-predictive profits, which are both non-predictive / fun markets and profits made after information comes out.

cc'ing @DavidChee here because there is some interesting discussions here

predicts NO

@TravisLuckenbaugh I don't think I understand what's the problem with the market... It looks like the probability going down as the month end gets close and the number of engaged users does not grow and then hitting ~0 as the final number is published.... So what's the deal?

predicts YES

@MartinModrak The market should have closed August 31st, 11:59 PM. After the resolution was certain, people with pre-existing piles of mana start betting amount amounts to run away with the market's payout. This market in particular was held open by its creator to enable this. Look at the trade history, the biggest winners (including creator @PC who got Ṁ28,521) were taking positions after the cutoff.

predicts NO

@TravisLuckenbaugh PC buys 4000 No shares after Aug 31 11:59. The market is at 1.7% before that trade, so the maximum they can get back is 4068 (I think it's actually 4056, based on the midpoint of before and after the trade, but I'm not sure). Later, they sell 7946 shares at 0.5%, costing them at least 40. Then they sell 990 at .9%, losing another 8. And finally 2968 at .9%, costing 26.

Someone can check my math, but it seems to me that they lost mana by trading after your proposed close time. Others may have made a profit, but I doubt anyone took as much as 100. Assuming PC was being rational, I'd guess the stat wasn't published until after their trades.

predicts YES

@Frogswap PC and others were commenting in the market about leaving it open. This is about as black and white as possible.

predicts YES

@Frogswap Also, your math is invalid. "PhilC sold Ṁ7,946 NO from 0.5% to 0.9% 20h" is not "costing them at least 40" it's an illegitimate gain of 7906 because at Sept. 1 3PM, the value of that security is zero. If he hadn't been allowed to hold the door open for himself and others not being able to recoup most of the bet money, he would not be up 28k to start with.

predicts NO

@TravisLuckenbaugh Can you link the comments you're talking about? Also I'm not sure what you mean about the math- I get that you're suggesting that PC was willing to take a greater risk because he thought he could potentially recoup up to 2% of the loss, but I don't follow how the value of the security is zero or how the gain is the total amount minus 0.5% rather than minus the amount spent on the position.

And I still don't get how someone can be the lead exploiter of the late resolution when they would have made more mana if they hadn't traded after when you think the resolution should have happened.

@TravisLuckenbaugh There were a lot of things going on yesterday. I hope we can have a productive discussion on this

@PC If I am understanding things right - you would like it such that markets close before the information comes out. Because once it does, it essentially becomes unpredictive

@PC I can empathize with this. For example, this is why I requested this market close before info went out because essentially all the profit would be made by the person who got to the information first, and I wanted to saw the real skill of the prediction, which is why I took a screenshot. (Note like in real markets, who gets the info first is valuable and rewarded)

@PC I can also empathize that smaller accounts can't execute this strategy because they don't have enough mana

Yeah I think it's good to end the market before the result is known when you can, but to speak to this market in particular it shouldn't be mandatory to do so imo.

@Joshua Finally, I also want to say that many people are making profit off of things that have occurred, and this doesn't reflect true forecasting ability. So, a lot of the higher earners on Manifold are those who spend a lot of time finding these markets.

I think as this may be less of an issue with more traders as potentially these opportunities get taken more, but maybe you have more markets to compensate

@PC That is all to say that you are pointing at some valid here.

I want to highlight a few of things:
1) It is helpful to get the info early - for example, Chris found a way to get the engaged #s faster, and people can trade off of that
2) I would rather have markets stay open because the issue for me is that I end up with large amounts of mana locked up. I would rather sell my position off for a small loss, so that I could use it elsewhere

3) These markets are competitive, so anyone can bid down to 0 and make "free money" as you said, but most of the time is it less than 1%, which even with 10k is only M100

I'm genuinely upset if you think I made the 28k because I left the market open...

predicts YES

@PC As I commented elsewhere, I think you're at minimum trying to be reasonable. Did you make 28k on that market that way? No. Did you make a non-zero amount? Yes. How much? You've pointed out yourself that Manafold makes figuring that out much harder than necessary.

predicts YES

@PC One thing I've experienced on markets "more fair" than Manafold: after I get an outsized amount of points / fake money / whatever often from lucky bet against the market going my way, it becomes much easier to make money to use the money I have to put pressure on markets.

My goto example here in prediction markets attached to some Twitch channels. The average market bet is a near coin-flip but the occasional outlier will have one probability much higher. Most reasoned partcipants would can guess this in advance. The returns on swamping those "closer to certain" bets resulted (for me anyway) in much higher gains as long as I was betting less than 5% of my total bankroll to prevent black swans from reversing my previous fortune.

This problem is IMHO even bigger for Manafold, but I'm not sure how to measure it.

More related questions