Skip to main content
MANIFOLD
What caused the change in the creators leaderboard?
7
Ṁ440Ṁ1.8k
resolved Apr 27
60%48%Other
15%3%
changed how it accounted for creators trading on their own market
15%10%
Bots are now excluded from the count
10%34%
bug related specifically to counting traders on unlinked MC markets
2%
changed how it accounted for alts
0.4%
Barak created thousands of alts
0.5%
[ignore]
2%
unique traders from accounts too old are now excluded

Resolves to closest. Since this is a shortfuse market, it may resolve N/A if too ambiguous. Please don't add an option that's a subcategory of some other option unless said other option is so vague it ought to resolve N/A. also add specific enough questions that some existing question isn't a subcategory. Like "a miscalculation" could involve a miscalculation within an MC market, or a binary market, or all markets at once, or some other thing, so is too vague

See context here: https://discord.com/channels/915138780216823849/918236290439335936/1230219138656370790

Market context
Get
Ṁ1,000
to start trading!

🏅 Top traders

#TraderTotal profit
1Ṁ31
2Ṁ23
3Ṁ20
Sort by:

I don't know what this refers to - is it about adding answers to others' questions? Because that's not specific to unlinked MC markets, you can also add answers to linked multichoice and the change presumably affects them too (you do get unique trader bonuses for answers you split off of "other" on someone's linked multichoice question, right?)

From the relevant commit that I see has been discussed on discord:

https://github.com/manifoldmarkets/manifold/commit/a96aec346a0e8b7819e90296a14f480d2f75c3de

Previously, the database query counted unique user–market pairs, the obvious, simple defintion of unique trader count. Didn't even look like it excluded bots or anything, though I suppose it's possible some other code was filtering them out later. But I would guess not. I know bots and API trades are excluded from the per-market unique trader count that you see at the top of each market, but I would go ahead and assume that the per-market unique trader counts just didn't add to the total unique trader count. Maybe it's closer to agreeing now.

Now, the total unique trader count is instead counting how many unique trader bonuses a creator has received. This is a different thing in several ways. Bot trades, API trades, trades on unsubsidised markets, answers to polls, don't pay unique trader bonuses. Answers added to others' questions do pay unique trader bonuses, to the adder but not to the creator, so that's why some people could have a higher count than before, if they added some really popular answers to some markets.

Also, bugs - there were a few days when everyone just didn't get unique trader bonuses a few months ago! So the new counting is gonna miss them too.

Pretty sure the speculation about "old accounts" or "small mana bets" being excluded are off base. The "created > $2" or whatever in the code looks to be filtering for bets after a given date. This is very like just because the "get unique traders" function takes a date as an argument so that it can only return unique traders since that date, in order to calculate the one-week, one-day and all-time unique trader counts (this is e.g what is used to calculate the "+X%" 7-day change stat you can see on your questions tab).

@chrisjbillington Thank you! Kinda tricky to resolve some of the partially correct options in that case. These seem to kinda apply:
- changed how it accounted for creators trading on their own market
- bug related specifically to counting traders on unlinked MC markets (though it doesn't seem to rly be a bug?)
- Bots are now excluded from the count

But mostly, Other: the count now is based on number of trader bonuses received.

I'm not sure what to assign to each answer.. hmmm. Closed so that it sends me a notif soon to resolve it when i am less busy

seems like partly MC counting separately, partly polls not counting, and partly small bets not counting?

seems like some of that isn't accurate, trade at your own risk before we know for sure! I will likely resolve at percentages

also a good guess. also mysterious that fuh's trader count would have gone up

yeah this is a good guess imo

bought Ṁ10 NO

@Bayesian but also why would FUH's total traders go up? could be that one of his markets popped off but i doubt it

bought Ṁ35 YES

if you decided to exclude this, no hard feelings =)

bought Ṁ30 NO

@AmmonLam yeah i think it's too vague for the market, will add clarification in description, you should uh sell, i can't N/A it. other ppl don't bet it down ok. alternatively give a more specific thing i can edit it to..

bought Ṁ450 NO

@Bayesian thanks. honestly I wont be mad even if I cant sell out early but thanks a lot for the ping, I did sell