Will manifold introduce a mechanism for retroactive market closure in 2024?
Dec 31

Metaculus has a mechanism for closing questions retroactively, such that predictions made after a certain point in time don't count. Presumably the purpose of this is to ensure participants are rewarded only for actual predictions, and not for "predictions" made after the outcome of an event became known, but before market closure.

Will Manifold introduce a similar mechanism this year? Anything that allows ordinary market creators to retroactively close a market, such that bets made after a certain date are treated similar to an N/A resolution, or anything substantially similar to that would count.

Functionality used only by admins in special circumstances does not count, but functionality used by admins as part of an understood normal functioning of Manifold does count. Basically it's a YES if one could create a market and realistically expect retroactive closure to apply, regardless of who actually performs it.

Relevant timezone for "2024" is US Pacific Time.

Get Ṁ600 play money
Sort by:

I am not sure how typical I am of the Manifold user base, but I would be wary of too aggressively using retroactive closures. The main reason is that I am on Manifold for fun and news trading is fun. In some sense, I use Manifold as "news gamification" - when I read the news I often think "there was a market about that, quick make a trade". Sniping a market after a news item feels good. Retroactive closures means there's less options for news trading and thus less fun for me and also less options to gain profit as a somewhat casual user.

Second risk is that except for some very clear cut cases (sporting event results, stocks crossing a price, ...), the point of news release is usually not the point where complete certainty about an outcome is available, although it may appear so in retrospect. A small fraction of news are later proven wrong for various reasons, so by sniping a market you are always taking that tiny risk and I think there's value in the market trying to price that possibility in. So I think any policy on retroactive closures should prefer choosing somewhat later close times than early ones (e.g. complete consensus of all sides rather a first newsflash)

With that said, I can see good uses for the tool, I just wouldn't want it to be used too aggressively with very early close times. It is also possible that the majority of the Manifold community wants to be more serious and less fun and that's also OK.

You mean my bot can fade the news all day every day, and get the slate wiped clean whenever it's wrong??

there goes my profit graph

bought Ṁ250 YES

v important for creators in a post-pivot economy. also a good look for sitewide calibration to outsiders. I heard some staff talk about wanting to do this at Manifest. makes me think it's in the planning phase, but my bet is prob optimistic

hell, as a trader who also often gets burned by my outstanding limits it'd be nice, too!

It’s important for only spending the liquidity subsidy on bets that happen before the event. Also the maybe the subsidizer should get to choose a prior probability so he’s only spending the subsidy on updates to his prior. They can sort of already do this with a smaller subsidy plus buying yes no, but that’s a less efficient use of capital

How do you do this and also remove N/A as a resolution though

Potential compromise is N/A should only require a moderator if you’re re-resolving something that’s been resolved for weeks. That limits the value at risk from a compromised creator account. Conditional markets shouldn’t require moderator intervention. Messing up the configuration of a market and cancelling it 1 minute later before anyone trades on it shouldn’t require moderator intervention.

retroactive n/a is also a risk for for instance arb bots, if one market does and another one doesn't

It'd also be nice to retroactiely close and then reopen markets. This would let you undo creator clariications that were against the guidelines and similar.

Extra note:


It required manual intervention with James but we rescued a broken market by manually performing this task. EVERYONE WAS HAPPY.

Which market was that?

Thank you for making this, but once again I have no idea how to bet.

How bad I want it: 99.9%

Do I think Manifold creators deserve this ability? 100%

Will they actually do it? No idea, they're already making a dating site and a politics site, I can't say what they will do.

Am I well-informed? I think I have a pretty good idea about what Manifold has been up to in the last few months.

Ian upthumbed the original suggestion thread, which is at least worth something.

Technically, I don't think this is particularly daunting. From my understanding, it will "just work".

I think it's over the current 28% I guess.