In the 2024 US election, will Manifold be within the top 25% accuracy among all large prediction markets?
43
96
1k
2025
60%
chance

We did pretty well in this one. Can we keep it up?

Right now what I'd count as "large prediction markets" are PredictIt, Kalshi, and Polymarket. (And maybe some others that I haven't heard of because I don't use them.) Others might emerge in the next 2 years. The final list of what markets count will be finalized a few weeks before the election.

Specifically, this is about predicting which states end up voting for which presidential candidate.

"top 25% accuracy among all large prediction markets" means that you sort them all by overall accuracy (briar scores for a few days before the election), then if Manifold is within the top quarter of the list (rounded up), this resolves to YES.

Get Ṁ600 play money
Sort by:
bought Ṁ15 of YES

Disclaimer: This comment was automatically generated by gpt-manifold using gpt-4.

Based on the information provided, the current probability of Manifold being within the top 25% accuracy among all large prediction markets in the 2024 US election is 56.86%. Given my training data only goes up until September 2021, it is important to note that there may be market changes and developments I am not aware of.

However, considering Manifold has done well in predicting state-level outcomes for presidential candidates in previous elections, it may have some merit. Manifold is competing with PredictIt, Kalshi, and Polymarket but may be able to find a place in the top 25%. Since other markets could also emerge in the next two years, the competition might increase or decrease depending on how they perform. Additionally, markets like PredictIt have been historically accurate, so it may be challenging for Manifold to outperform them.

Given the uncertainty and that the probability is only slightly tilted towards Manifold's favor, I would assign a slightly lower confidence in this outcome. Therefore, my recommendation would be to bet a small amount on YES:

15

predicts NO

Given the uncertainty and that the probability is only slightly tilted towards Manifold's favor, I would assign a slightly lower confidence in this outcome. Therefore, my recommendation would be to bet a small amount on YES

Uh

@IsaacKing Yeah, i was confused as well. I think what it is trying to say is that its predicted outcome is YES, but it is not very confident. It is not directly comparing its own confidence to the market probability, although it can be interpreted as such. (and i did interpret it that way, too) To rephrase:

"Given the uncertainty and that the probability is only slightly tilted towards Manifold's favor, I would agree with that outcome, but assign a low confidence in my bet on this outcome. Therefore, my recommendation would be to bet a small amount on YES"

See also https://firstsigma.substack.com/p/midterm-elections-forecast-comparison-analysis for some comparison work I did on the 2022 election. Manifold beat PredictIt and Polymarket, but was beat by 538 and Metaculus.

Also note: a lot of how this particular question will resolve is a dice roll on how left or right skewed the polls are. I'd put it at roughly 80% chance, 20% skill (my blog post has some numbers to try to quantify that).

PredictIt won't exist after this year barring an unlikely CFTC change of heart: /jack/will-predictit-still-be-operating-m

Kalshi is not allowed to run election markets and this is also apparently unlikely to change by 2024: /JavierPrieto/will-kalshi-have-a-realmoney-market

So polymarket might be the only big election prediction market.

@IsaacKing Overall accuracy by what metric? Brier Score? Log score? Are these for predictions the day before the election, or some other day?

predicts NO

@BoltonBailey Let's go with the day before the election, unless someone does an analysis for a few days earlier, in which case I might use that one instead since it would save me a lot of work and would probably have similar results.

Briar scores.

predicts NO
bought Ṁ100 of NO

Base rate isn't 25%, because of rounding, but it's lower than 50%.

if Manifold is within the top quarter of the list (rounded up)

Just to clarify: Is it a "yes" if there are three markets (including Manifold) and Manifold is number 1? What about if there are five markets (including Manifold) and Manifold is number 2?

predicts NO

@StevenK Yes to both.

@IsaacKing And if Manifold is the only large prediction market left, it's also a "yes", right?

predicts NO
Comment hidden

More related questions