I'm sorry if this market became such a mess. Most markets should account for methodology changes when using external resolution source.

This one resolves YES, as it most probably should. The 37.5 resolves N/A.
Thank you for game and comments, see you in other markets!

An argument that we haven't taken into consideration is the point of significant digits. This is important because significant figures detail the extent to which numbers can be considered accurate. Note, surveys /polls already often reflect a high level of uncertainty and fivethirtyeight's whole selling point is to reduce this error. Considering the original question was based on three significant figures, and five thirty-eight's page also only reports three significant figures, out of accuracy I think it would be make sense to keep it at the closest rounded number - i.e. never having dropped below 37.6

@EricBurns Agree. Since the rules said nothing about looking beyond the web page and the single decimal place, it would be unfair to use the underlying data. In my view, both markets are no.

I can see the spirit of the market making this a yes since we have been under 37.6 in both datasets. However, the letter of the law here said nothing about consulting underlying data, and quite clearly made the point--using a screengrab no less--that it would be based on what the web page says, not the .csv file. For that reason, I would argue that both markets are currently no. I have divested from this market because I can understand if @itsTomekK wants to resolve based on the spirit of this market using the underlying data.

The 37.5 market is currently a no no matter how you look at it. Neither the website nor the underlying data (using either Silver's or Morris's data) has ever gone below 37.5.

@ScottSupak @itsTomekK
This is what I mean, basically we have not been trading on the same charts in a sense.

The 37.6 Market Has been trading off of the new chart style the whole time.

The 37.5 Market Has been trading off of the old chart style in the description with the new chart data being what is reported in the comments and seen by predictors.

You can see it in the differing of numbers from the SAME DATES.

For this 37.6% market, the description shows 40% on June 30th 2023

For this 37.6% market, as of TODAY 12/19/2023 it shows shows 40.3% on June 30th 2023

For the 37.5% market, the description shows 37.5% on July 21st 2022 (notice the chart is different color also):

For the 37.5% market, as of TODAY 12/19/2023 it shows 38.4% on July 21st 2022

@itsTomekK

Actual All Time Low From Same Time Period 37.9%

The December 2nd and December 7th 37.6%

Obviously this is using VISUAL tool provided in the description and not GITHUB DATA, which is not provided in the description but more accurate.

@itsTomekK What Date Did You Get Your 37.6% All Time Low From?

@SirCryptomind I guess it was all time low when i created this

@itsTomekK See my comment about the 538 charts posted on both markets, my explanation is a bit better here.

@itsTomekK it's looking likely that we won't hit another low. Can we get ruling if the previous low doesn't counts for the resolution, because it did not fall below 37.6 in the graph (and thus NO for now) or if it counts because of the more precise data outside of the 538 website etc?

Thank you for bringing that up @Weezing.
I honestly don't know what would be fair now, since not only there are two markets, but also the methodology has changed.

It is the all-time-low
It is below 37.60 by fine data
It is not below 37.6 on chart... but with previous methodology it would?
In case of a tie (37.6 vs past 37.6 (now 37.9)) fine data would be used

... and then there is second market, which also should get ruling (what should it be?)


Your comments are appreciated

@itsTomekK This market will resolve to YES, if at any point of 2023, the Approval Rating for Joe Biden falls below 37.6%

I think that’s pretty clear resolution criteria.

@itsTomekK Everything you said seems to point to YES. Even what Jim said implies YES, because it did indeed fall below 37.6%, it just got rounded up. I think the other market might be controversial, but not this one.

I agree with shump that this relatively clearly resolves yes and the 37.5 one is more controversial.

I continued buying yes after it hit the recent low both because the exact value was lower than 37.6 and the low has been hit using the current methodology.

I feel like this is a strong case because the title and description both point towards yes.

I didn't bet in the 37.5 market because it seems less clear. While I think think the more specific criteria in the description should usually trump the title, I think the methodology change warrants a criteria change.

I haven't looked at the information surrounding this site so I don't know if the needed information is available to us, but I think using the old methodology to find the current low would be the best way to resolve.

If that isn't doable, I think resolving to what the current data says also makes sense.

But, I think it also matters when the methodology change happened. If the numbers changed a while ago, resolving (i.e. Some amount of time before the value hit the recent low), it could make sense to resolve NO (if it still doesn't go below 37.5) because people were trading with the old numbers in mind. Though this makes me less comfortable because it's way less black and white.

P.S. I started writing this thinking 37.5 should resolve NO but ended thinking YES, lol

@MichaelLewis to be clear, I think the markets should be judged in the same way, but I assume the only way this market resolves NO is if it is decided that rounding is important.

Both Markets Should Reflect The 1 Lowest # ; Not 2 Separate Markets With 2 Separate Lows. If 1 Resolves YES as I believe it should, the other should also.

I will not be making a decision on this as a Mod, but I will state my opinion which may not align with others from Manifold , including other Mods.

I DO Hold A Position In Both Markets

@itsTomekK thanks, here are my thoughts on this matter:

In my opinion what is important are the original resolution criteria written in the post, because that was what the most people were trading on presumably.

"This market will resolve to YES, if at any point of 2023, the Approval Rating for Joe Biden falls below 37.6%
resolution: 538, https://projects.fivethirtyeight.com/biden-approval-rating/"

Same description is for the other market with different threshold. I think it clearly says the threshold and method for determining the current value (the number displayed in the graph on the website).

That would mean both markets are a NO in my opinion. Definitely the 37.5 because it did not fall below it at all. However I agree this one is more controversial, because in the external data it technically fell below it. But again, the description states it will be determined according to the 538 website, where the number of displayed as 37.6, which means it is a tie and the the conditions for YES resolution is not met.

The methodology change is unfortunate, but I don't think it matters that much. Again the important should be the number of the 538 website, which was stated resolution criteria.

I hold small NO position on the 37.5 market, because I though that one is very clearly NO a thus is safe (and seems like most other traders do too) and I sold my position here earlier, because I though this one might be controversial. That's why I pinged you to make a final decision, because it seemed people in this 37.6 market were mostly trading on speculation of what you will decide after it is closed and that will always end badly.

@SirCryptomind I definitely disagree that both have to resolve the same way. Each one has different resolution criteria clearly stated in the description, and thus might not be resolved the same way.

I think the correct resolution depends on @itsTomekK's intent. The issue that that the original title (all-time low) and the description contradict each other. Tomek should resolve based on which of these was the main intent and spirit of the market.

@Shump I agree and that's why I'm out of the 37.6 market. I like @itsTomekK but I'm not interested in betting on what he'll decide!

@Shump I dont think there is a contradiction between title and decription. Title just says approval all time low which is a generic term and it can be measured in multiple different ways. The description specified how it will be meassured specifically: the market will be resolved based what 538 website shows and the selected threshold ("< 37.6").

Edit: but you could say there is contradiction between purpose of the question (will biden become less popular than he was before?) and resolution criteria. That's consequence of it being extremely close and some unfortunate methodology changes a while back. But I think title and description were fine, since it was hard to imagine something like this would happen.

Wait, the old low from 22 is now showing 37.9%. I guess they made a retroactive methodology change? @itsTomekK Shouldn't this resolve YES then? I mean, it IS an all-time low. I don't think that's really fair for current year data otherwise, because the entire thing got adjusted up.

More specifically, the current data file shows a low of 37.582 this year, and 37.871 last year (on 13/07/22)

@itsTomekK

@SirCryptomind Damn, you are probably right about this one. How about 37.5? It's still all time low, but not below this number

@itsTomekK If the meaning was 'the market resolves yes if the bold line display ever falls below the previous lowest point of the bold line display, there would be no difference between the two markets. I think the most obvious reading is 'the market resolves yes if the bold line display ever falls below [37.6 // 37.5, depending]. Given that I think 37.5 market should count as NO assuming the bold line never hits 37.5 between now and the end of the year, and 37.6 market should be counted as a tie, since the bold line hit exactly 37.6.

The resolution criteria says "In case of a tie, detailed data to calculate averages will be used to resolve." This is not strictly clear but I assume refers to the fivethirtyeight downloadable 'trend line' data, where we see that the lowest value of approve_estimate was 37.58169144640985, which is lower than the stated value of 37.6. Even if you decide to compare the precise value to the previous precise lowest value, it's still lower. (37.62198863993981 )