Which state will be the tipping point in the 2024 US presidential election?
➕
Plus
124
Ṁ39k
Nov 13
86%
Pennsylvania
5%
Michigan
4%
Wisconsin
1.1%
Georgia

This market resolves to the tipping point state in the 2024 US presidential general election.

Colloquially, this is the state that “tips” the election to the winner, allocating states in order of decreasing margin.

In a bit more detail:

  • Let “winner” be the candidate who wins at least 270 electoral college votes in a US Presidential election.

  • Let a state’s “margin” be the difference in votes between the winner and second place finisher, divided by the number of votes cast in that state.

  • Then sort states by decreasing margin, and begin counting electoral votes by state. For some state, after its votes are counted, the winner will have at least 270 electoral college votes: this is the tipping point state.

  • See Nate Silver’s article for an illustration of tipping point states.

  • If there is no absolute majority in the electoral college (eg, a 269-269 tie), the market resolves to None.

  • Manifold will resolve this market when the popular vote counts released by the Associated Press are comprehensive enough to make this determination, i.e., after all states have been called, and the popular vote margins are known within sufficient margin to determine the tipping point state.

  • It is possible for the tipping point “state” to be a district of Maine or Nebraska, because these states allocate 1 electoral college vote per district in addition to 2 for the entire state. In the preceding description, “state” means “state or district: NE-1, NE-2, NE-3, ME-1, ME-2.”

Get
Ṁ1,000
and
S3.00
Sort by:

Same idea but for the house:

bought Ṁ500 Pennsylvania NO

If you don't know anything about the ranking of margins, the prior is roughly proportional to the number of electors, right? So putting PA (19 electors) at the same probability as for example MI+WI+NC (41 electors) seems like it implies a substantial amount of information about how the margins in those four states will be ranked. I haven't really been following, but I'd be kind of surprised if we have that information.

The relative ordering of states in 2020 and 2016 provide a strong prior for the order in 2024. MI/NC have been further from the median than WI/PA, only by a couple percent but that's not a small amount in very inelastic presidential elections.

Related Market:

Although it would be the most unlikely outcome, it's strange that every possible option besides DC is listed here

bought Ṁ10 Michigan YES

Michigan seems undervalued here

Existing related markets:

@ManifoldPolitics The second of those is a better market than this one, in my opinion. Including the improbably states actually makes the question worse. It slows down the UI, and degrades the usefulness of No orders because of the problem of excessive capital requirements for No bets in linked multiple choice markets.

The first one is so bad that it should not have been included in the comment. Nothing at all against the question author or how they wrote it, it's just an old question and the parimutuel structure, which is a wholly inappropriate structure for this sort of question where information will trickle out over time and things gradually become more certain. Late betters get a significant advantage. (As opposed to, say, parimutuel betting on a race outcome where closing betting before the start of the race mostly fixes the problem.)

@EvanDaniel even the first one has some amount of trading on it, and perhaps more history, and i don't see the harm in including it?

Re all the excessive options, your points make sense to me but I wanted the market to be comprehensive. We'll push an update at some point to update the sort of the options, and hide the small ones, and if it gets more trading my understanding is that the small probabilities some states have will continue to go down. If the other market stays more popular than this one forever, then it will update my beliefs

@barak

even the first one has some amount of trading on it, and perhaps more history, and i don't see the harm in including it

What fraction of current Manifold users do you think know that parimutuel markets exist on Manifold? Understand what they are and how they're different from the linked MC format? Would prefer to bet on that format vs the linked MC format? Know that it's parimutuel before they click the link?

Including it is a UX problem. Including bad options is bad UI/UX design.

I wanted the market to be comprehensive

I think that should be a distant second priority to good UX. You could have used the "other" option and added more states if that stayed persistently above the risk-free rate or something.

Manifold doesn't currently have UI, market math support, or API primitives to run the comprehensive version correctly, and fixing the sort and display issues isn't adequate to address that.

Are you at least planning to benchmark performance under heavy load and compare this one vs. the smaller one under reasonable assumptions? What's your performance target for the UI? Have you personally tried to trade in any of the highly popular markets that had lots of options recently during periods of heavy activity?

If the other market stays more popular than this one forever, then it will update my beliefs

I hope you're also planning to think critically about which one would have been more popular if they got equal advertising and "official" support and so on.

@EvanDaniel

2c. What's your ballpark guess for how that performance comparison would turn out? is it linearly worse, or exponentially worse?

3. That's very fair. I guess I don't have a strongly held belief that markets need to be comprehensive in their choice list. I guess then, I will await more information, vibes, and benchmarking and have already slightly updated my beliefs

Thank you for your clear and persuasive commenting/energies

@barak I don't have a good guess about the benchmark results. Weak guess, having only interacted some with laggy markets, and mostly just watched them and watched the discussion: super-linear, but not nearly as bad as exponential. Quadratic wouldn't surprise me, nor would quadratic with a relevant linear component across the relevant sizes. (That is, I think n is small enough that O(n) approximations might not be the only relevant thing.)

And to put all my complaints in context... I think this is a very cool project and I'm very glad someone is prioritizing it! I am also in favor of making mistakes along the way, trying stuff out, ending up with redundant markets or not-quite-ideal markets or whatever. As long as we're learning from things and understanding what's working well and what's not and improving.

© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules