Any form of randomness that triggers a market close with out need of a bot or use of an API key, implemented before 2024, results in this market resolving YES.
The prohibition on the need of a bot or API key is meant to exclude third party involvement beyond observing a third party source of randomness. The random close should be something that the market creator sets up at creation and can be in addition to or in lieu of a fixed or outside closing date.
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ113 | |
| 2 | Ṁ37 | |
| 3 | Ṁ27 | |
| 4 | Ṁ26 | |
| 5 | Ṁ11 |
People are also trading
Was thinking of something similar with /Stralor/will-manifold-add-a-market-closing-2b6bdb889c03 and /Stralor/will-manifold-add-a-market-closing
I have the scheduling logic here, but wasn't sure the front end work was worth the time. https://github.com/manifoldmarkets/manifold/pull/1739
@ian It looked like every hour there was a 10% chance of closing for the linked pull request? so after the "top of the hour" there would be an hour of guaranteed trade available? This would count, but it feels semi-random since after the script runs there is an assured unit of time during which the market would not randomly close.
If the random close was evaluated each time there was a trade it would remove any fixed safe times, however I assume that the incentive is usually to the be final trade executed so if the randomness only had an opportunity to close the market then it could incentivize people to spam trades to try to force a close. So maybe the market might have closing range (close after, close by) used to configure a randomly chosen secret "actual close" time and then each trade after the "close after" time randomly modifies the "actual close time" by a scalar, that can be greater or less than 1, of the time remaining to actual close or by some random number of minutes added or subtracted from the "actual close" time. Such a random close market could by a random walk method walk outside of the original (close after, close by) range but would tend to close at the first randomly chosen secret "actual close" time within the (close after, close by) range.
@ShitakiIntaki To resolve the market with 10% probability per hour, but not at the top of the hour, you could draw from an exponential distribution with a mean of 10 hours, to obtain a time interval, counting from now, after which the market will close. Store that timestamp and close at that time. If you lose the timestamp because the code restarts or something, no need to put the scheduled random close time in a database or anything, you can just compute another one in exactly the same way - the statistics of outcomes will be identical.
You can draw from an exponential distribution by taking the natural log of a a number drawn from a uniform distribution between 0 and 1, i.e.dt = - prob_per_hour * 3600 * log(rand())close_at = now + dt