Will we be able to trust that our portfolios are correct by the end of 2023?
25
176
แน€470
resolved Jan 1
Resolved
NO

Right now Manifold consistently encounters bugs that cause people to incorrectly lose or gain mana/shares, or have their balance/shares not reflected properly in graphs or other displays. This market resolves based on whether it seems that's no longer happening frequently at the end of the year.

Get แน€200 play money

๐Ÿ… Top traders

#NameTotal profit
1แน€347
2แน€208
3แน€15
4แน€13
5แน€13
Sort by:
bought แน€500 of NO

Seems almost guaranteed that this market is going to resolve NO. Style points to anyone who manages to get a balance/profit bug to occur for that resolution.

predicted NO

Some more bugs that have happened:

1.

In late October, Manifold accidentally gave some users free mana by giving them double loan payouts on one or two days (I got two), with the extra loans counting as profit and having no corresponding liability. This was not reversed (correct reversal would require loan amounts were actually recorded somewhere, and actual records were...inadequate). My all-time profits and balance are about 23k too high because of this. Here is the market about whether it would be fixed:

/chrisjbillington/will-i-have-to-pay-manifold-back-th

And here's the discord bug report about it:

https://discord.com/channels/915138780216823849/1167747252765343786

This was never rectified, I never had to pay it back.

2.

On November 17th during the Starship launch, I encountered a bug that allowed me to bet on and re-resolve a previously-resolved answer on an independent multi-choice market. I accidentally did this with a script I was using to resolve other answers. James did database surgery to make things mostly right, but my excess profit was removed with a negative managram, which did not decrease my profit metric.

Discord thread here:

https://discord.com/channels/915138780216823849/1175435915083128942

And the fine here:

(and a reward for reporting the infinite mana glitch that this bug turned out to be)

So my all-time profit is a further +14k too high because of this.

3.

Leagues profit calculations are currently subject to the following bug:

  • Buy "other" in a multiple choice market in one month

  • In the next month, someone adds an answer, splitting "other"

  • The redemption bets this creates count as bets this month for the purposes of leagues profits, even though you didn't make any actual bets.

4.

Leagues profit breakdowns don't list all markets for some reason, I'm not sure why:

https://discord.com/channels/915138780216823849/1186920170598432888

Most users are only missing 2% or so of their markets but the one discussed in this link is missing like a third.

5.

There is a daylight-savings bug that means depending on your timezone, the bets displayed in your leagues profit breakdown are either missing one hour, or include bets from one hour more than they should:

https://github.com/manifoldmarkets/manifold/issues/2242

6.

User portfolios only show 5,000 markets, maximum. Check your browser's requests when loading firstuserhere, botlab, or acc's portfolio page. You'll see only 5000 markets being loaded. As an example, load @acc's porfolio, view "all" markets, and sort by oldest. I think you'll agree there should be markets there that acc bet on longer ago than 26 days!

7.

This one likely isn't verifiable, but is useful if you trust @Conflux, who had an N/Ad market showing profit for a time when it shouldn't have been (since fixed):

https://manifold.markets/Conflux/will-the-bugged-multiplechoice-mark

8.

Loans counted as negative profit for a day or two recently, here's the bug report on discord:

https://discord.com/channels/915138780216823849/1187514428359442483

9.

Profit metrics didn't update at all for something like 36h a few days ago. So the metrics were wrong, in the sense that they were more out of date than usual and than intended:

https://discord.com/channels/915138780216823849/1187236459384672306

10.

Austin manually incremented the balance of the @Manifold account a few days ago, in order to give it mana to transfer to ninthcause as salary, and maybe some other things, I'm not sure. Anyway, it was done incorrectly and counted as profit, leading to a 20 million mana profit spike, which went away due to austin manually fixing it:

The Manifold account temporarily ended up at the top of the leaderboard as a result:

discord thread about it:

https://discord.com/channels/915138780216823849/1187509137953783818

11.

A bug here on discord reported by you (Isaac) about a month ago, that was fixed, about profit displaying incorrectly in some circumstances. As described by Eliza: "When an unlinked MC answer is resolved, it is currently using resolutionProbability even though the figure is incorrect"

https://discord.com/channels/915138780216823849/1178489248500940811

12.

Profit spikes continue to be regular occurrences:

https://discord.com/channels/915138780216823849/1155571113280147487

13.

A bug reported about two weeks ago (I assume fixed, not sure) about the loan amount for unlinked multiple choice markets being displayed incorrectly on the user portfolio page:

https://discord.com/channels/915138780216823849/1185161332941672488

14.

Displayed profit for unlinked multichoice markets was just wrong for some time. It now looks correct, for the markets I can see, but it was wrong as recently as the height of the Sam Atman drama, where the profit shown for the "why was he fired" market was meaningless in one's portfolio, but correct if you viewed your profit after clicking through to the market's page. This one seems to have been so accepted that I can't even find a specific bug report for it now. My feeling is that there were so many bugs around those markets that it kind of was just expected lol. But it seems to be fixed now.

15.

The display bug mentioned in a below comment about BTE's one-day profit in a certain market being wrong.

---


Some of these things probably don't directly apply to this market (the leagues profit stuff, possibly), but nonetheless decrease my trust in my portfolio being correct. And it's definitely true that my all-time profit is wrong, currently, so I don't trust that!

bought แน€10 of NO

Collecting loans now shows up as a loss on your portfolio. All profit numbers are completely meaningless now. If this doesn't get fixed by EoY, we will have even less trust in our portfolios than we did at the beginning of the year.

predicted NO

@PlasmaBallin Not anymore, they fixed it the other day, and corrected the historical inaccuracy. It was only wrong for less than 48h I think.

But even though the incident is resolved, it's still evidence for NO on the question of whether it "seems that's no longer happening frequently", where "that" is portfolios being displayed incorrectly.

predicted NO

@chrisjbillington Are you sure it's fixed? I still see big negative spikes on my profit graph right at the time I collected the loan, and they don't seem to have been corrected for.

predicted NO

@PlasmaBallin It was fixed for me:

Spike down, and up again:

For you there's a spike down and up again, and then a second one down:

You reckon that second one is the same bug? It's true that it's about the same size, and yeah, I don't see a M1500 loss anywhere in your recent trading history. Hm.

predicted NO

@chrisjbillington I think I figured out what the second spike was from (It wasn't a bug, but was caused by this market: https://manifold.markets/Ernie/will-the-main-lk99-market-ever-reac?r=UGxhc21hQmFsbGlu). So I guess the loan bug has been fixed, then.

To clarify, this market is about bugs that cause displayed numbers to be incorrect. It's not about misleading design choices.

bought แน€220 of NO

@IsaacKing Yes. the "one day profit" zeroing out at midnight pacific time (or UTC-7, or whatever it is) is a poor design choice and wouldn't count by itself. The below screenshot I think is just showing a bug though. Do you agree?

@chrisjbillington I haven't checked the numbers independently, but if you're correct that most of the overall movement is just from that one market, that seems like a bug to me, yeah.

predicted NO

@IsaacKing Yes, that's how it looks. You can see the current value of his shares is 36k, and he didn't buy 36k worth today (ss below) - most of that value is an increase in value today of shares bought previously. The total profit column showing +27k is reflecting this, and the one-day profit should be similar (likely a bit higher - he would likely have made an overall loss in this market prior to today, though I haven't checked for sure).



predicted NO

@BTE's profile is currently showing a +40k one-day profit, which is mostly due to the movement in /BTE/will-there-be-a-military-conflict-b

However his portfolio shows a one-day profit on that market of only แน€1.2k, this is obviously incorrect.

Almost all of the value of the shares BTE holds in that market are from movement in the past few hours. The time between then and now doesn't overlap with midnight pacific time or anything like that that might affect the 1d-profit calculation intentionally, it's just incorrect by any definition:

bought แน€220 NO from 57% to 35%
sold แน€239 of YES

Well, I'm definitely getting weird values right now. Showering major loss on the graph when not accurate. I think the sum is actually correct but the graphs are all weird

predicted NO

@Abraxas Yeah there's a bug where loans are counting as negative profit currently.

https://discord.com/channels/915138780216823849/1187514428359442483

"no longer happening frequently" is a very different resolution criteria than the headline.

@EugenGrue How so?

Related markets