You people are insane lol

@JimM9593 did you see the Dixville results? 💀

@JimM9593 only people not called jim

So if I read "the" poll correctly 736 voted for either harris or trump, with 380 for harris. So conditional on that the poll is as high quality as claimed..

Applying Confidence interval https://en.wikipedia.org/wiki/Confidence_interval

gives 95% chance the harris loses by 2 points to win by 5.

99.9% chance between loses by 4.5 points to win by 7.6 points.

Meaning highly unlikely trump with by >5 points. (or the poll is wrong)

There's some arguing in the comments about whether Trump needs to win Iowa by >5 points to win the election. So I made a made a market based on those 4 outcomes:

/Ziddletwix/will-trump-win-iowa-by-5-points-and

This could easily be arbitraged with the general election market, there is no world where Trump wins Iowa by less than 5 and he wins the white house.

@HenryRodgers probably, but the complementary is not necessarily true.

He could easily lose the general and still be >5 points in Iowa

@hidetzugu Yes, exactly, which is why this market can so easily be arbitraged since it should be greater than or equal to the chance of Trump winning by definition, and it currently is less than that.

@HenryRodgers

https://manifold.markets/ManifoldPolitics/will-trump-win-the-2024-election#yoh4sixb5z

@HenryRodgers

there is no world where Trump wins Iowa by less than 5 and he wins the white house.

??? Of course that could happen lol. States realign at different rates each election cycle. If I made a separate market for that outcome would you bet on it.

@FergusArgyll Your core assumption is that anything about the 2024 early electorate can be gleaned from the 2020 early electorate, which (at least I believe!) is inherently flawed. There is a point where pure statistics ends and data science begins, and accounting for things like ballot return temporal depolarization is one of those. We’ve only got two days, just have patience and don’t try to make statements on the very loose ground of early vote partisanship.

@Ziddletwix I don’t really have much mana left in my balance that isn’t already in a different market, but sure yeah I would bet that market. It can’t be conditional though, I don’t want to waste my liquidity on something that will be NA’d.

The point though is that demographic shifts are largely universal across states, and the demographic realignment necessary to produce Republican backsliding in IA would hand the rust belt to Dems.

@HenryRodgers Oh, it's definitely flawed.

I try to adjust for that by taking the average movement from 2020 - 2024 and seeing which states deviate from that the most.

It's definitely not very accurate, nevertheless you do see that Iowa seems to have a different voting profile then many other states

@FergusArgyll I still don’t believe that comparing states to each other can get you much either. Countless laws and rules have changed since 2020 on early voting (shortened windows, differing eligibility, etc) and they’ve all changed differently in every state. We’ll find out if you’re right in two days, but I honestly think putting too much faith in early voting is just trying to read the tea leaves.

@HenryRodgers we barely disagree on the effectiveness of early vote analysis.

I do disagree w "if he loses this then definitely..."

Florida & Ohio were bellwethers for years, Trump lost while winning them. States change all the time (VA, colorado used to be swing states)

@FergusArgyll I'm sorry but this is so surface level I'm not entirely sure how to respond. Florida & Ohio's shifts are products of their demographics, and their shifts are visible across the country in how their demographics have realigned. Floridians and Ohioans do not magically vote different from people living in other sates. Ohio for example, was the victim of rural Midwestern voters shifting hard to the right over the last few elections, however it has become more of a solid red state than the rest of the Midwestern swing states because it has a much larger rural population (compared to a state like Michigan, for example) and lower turnout in its urban areas. These are all measurable trends, I understand you did a statistical analysis with code (which is great!) but there are much deeper layers to this and just saying "states change all the time" is just saying that you are uncertain, not that you are right.

@HenryRodgers we'll find out in 2 days, good luck!

@HenryRodgers well if you're out of mana fair but we can see how it shakes out for other bettors here:

/Ziddletwix/will-trump-win-iowa-by-5-points-and

@Ziddletwix I put a limit for 10% with all my mana, we'll see if anyone takes it. I think the actual chance is well below that but the return isn't really worth it unfortunately.

@HenryRodgers sure, but 10% is meaningless—can't imagine anyone would think it's higher than that? the original statement was "there is no world where Trump....". even if it's 4%, that language is way too strong. and when comparing different election outcomes, rounding something that's a 2-4% chance to 0% is a badly losing play.

@Ziddletwix I do believe that the chance is well below 2% but it's not really worth it to bet it down to that when there's hundreds of markets on here with essentially 100% chance of happening that are currently at ~95% odds or so.

@Ziddletwix What do you know, my entire limit filled! I now have 0 mana. Guess bettors here think it was a bolder take than you thought.

@HenryRodgers yup turns out we're closer to agreement than some then, given <10%.

that being said, the probabilities in that market consistently don't make sense—the marginals are often way off from the linked markets (especially P(trump win), which should be pretty fixed). so i'm not sure it reflects true credence, more that conditional markets like that are hard to make accurate and people are placing some pretty bad bets in it. (i will be continuing to take NO on that option)

@Ziddletwix Have you changed your mind? You've got some tempting limit orders there... :)

@HenryRodgers Mostly just liquidity. I don’t think it’s so much <<< than 10%, at least for something I haven’t tried to calculate precisely, and I’m far by the biggest no holder, would rather free up some mana once it’s in the right ballpark

Especially bc in that market the other probs were often more obviously off

@Ziddletwix That's fair. I'm not really confident on whether or not the Selzer Poll will be accurate, just that I don't think Trump has a reasonable chance overall if he's winning IA by <5, so I'm probably done betting on that market. If I had to guess, I think 50/50 here is way too bullish on the accuracy of polling, but I'm not really confident enough to bet against it.

@HenryRodgers Yeah I think the fact that the P(trump win) in that market keeps adding up to <<50% is the more glaring issue. I think that’s mostly just that the first option is underpriced but conditional markets can be pretty inefficient, it’s tedious work to get it right for small gains

@HenryRodgers of course we will have to wait and see, but, I think the Selzer poll will be shown to be a pretty inaccurate outlier, while it would not surprise me to see that Trump's overall performance in Iowa is slightly down from last election. I do think that there's some truth to the idea that there are certain demographics (of mostly women) that are motivated enough by the changes to abortion law in the state of Iowa that it will negatively impact Trump and the rest of the GOP downballot races in the state, but, given Iowa's general trend towards the right over the past 20 years or so I think it will be pretty close to +5 for him when we get the final results.