I'm experimenting with a way of incentivising somebody to come up with an answer that should be knowable but might be tedious to work out. I'm interested in this particular question because I've recently been running my balance pretty low betting in LK99. Sometimes I get annoyed that I run dry and then the probability changes and I wish I could buy again. Is everybody like this or are other users more prudent?
This market will close once somebody posts in the comments a convincing answer to the question. Assuming their answer is a number between 0 and 1, I will resolve to the corresponding percentage. If nobody posts a solution, the market closes in roughly 28 days at a pseudorandom time, resolving to the percentage the market is at at that point. If I suspect manipulation (despite the pseudorandom time), I reserve the right to delay closing.
What counts as "a convincing answer"? I think somebody probably needs to write some API code that calculates users' balance-to-net-worth ratio and then averages over users. For this question I'll ask for answers to be weighted by users, not by mana. Ideally they'd share their code.
Obviously, the answer will vary over time. Ideally the best answer will account for this, explore how erratic or consistent it is, and give an average over an appropriate time interval (where "appropriate" will depend on how erratic the ratio is).
If somebody makes a convincing case that the ratio is so erratic that you need to average over a longer period of time than there is until the market closes, I will extend the close date. (Although I suppose you'd also need to confirm that it's not possible to go "back in time" through the API.)
If people make contradictory but seemingly convincing answers, I will average the answers weighted according to how convincing I find them.
Every time I say "convincing" above, I'm the ultimate judge of that. If I'm not sure, I might leave the market open for a bit, say in the comments how convincing or otherwise I find it, and invite other commenters to weigh in.
If you have an estimate (ill formed or well formed), I encourage you to bet in the market to reflect that. If you have a full solution, feel free to do as much insider trading as you like before posting the solution (but make sure you don't fall foul of the close date). Or if neither of the above, feel free to just go crazy and learn some market dynamics, guys. You'll be helping us by subsidising people to find the solution.
I'm concerned this question might be a bit too easy. If I think it has promise I might experiment with more difficult questions. If anybody has any suggestions to improve the format, I appreciate them!
I won't bet in this market.
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ4 | |
| 2 | Ṁ2 | |
| 3 | Ṁ0 | |
| 4 | Ṁ0 |
People are also trading
Damn, I forgot about this market. Seems like a failed experiment, partly because I didn't manage to come up with a clear decision on averaging and truncating, and partly because as @PatMyron says, net worth isn't easy to find out.
Thanks for taking part, those that did.
I manually sampled a couple dozen using the league pages and got 1.24. Some people have very low or even negative net worth while still having a significant balance. For instance, @Chllucas had a ratio of 18.3. It's possible someone could have a net worth of 0 or 0.01 maybe. That would have a big effect on the final average. Most users seem to be between 0 and 1, maybe the values outside that range would need to be truncated or censored before averaging.
@travis damn, that's really surprising. I considered putting in more caveats about ratios outside the range but thought it wasn't going to be useful, and the description was long enough already.
I should probably wake up a bit more before making a final decision, but some thoughts:
It seems unfair to completely truncate values greater than 1 before taking the average. If there's a bunch of people greater than 1, it seems fair that the more "greater than 1" they are, the higher the result should be.
The fact that the range is mostly 0-1 but has some outliers significantly above 1 suggests that a geometric mean might be more reasonable than arithmetic. Trouble with that is a single user with balance 0 ruins the whole thing.
Negative values are a pain. It feels intuitively that there's a big difference between somebody who has positive bet worth/negative balance and subsidy so glad negative net worth/positive balance. And there might be some people out there who have negative of both and they'd show up "normal"!
Maybe truncating all negative values is reasonable, but dealing with values >1 is challenging.
Like I said, I'll think about this more after I'm more awake and try to salvage it. Thanks very much for pointing it out!