Using Monday's before the election probabilities, will Manifold probabilities "feel accurate" in retrospective?
➕
Plus
7
Ṁ2202
Dec 10
15%
chance

I'll use a three-pronged approach, using the probability of victory of the candidate with best probabilities on the Monday prior.

  • 80+% probability: the candidate wins the election. even if you change the winner vote share in each state by 1.5 percentage points, giving it to the 2nd candidate, the candidate still wins.

  • Between 60 and 80% probability: correctly calls the election. even if you change the winner voter share in each state by 0.75 percentage points, giving those votes to the 2nd candidate, the candidate still wins.

  • Between 40 and 60% probability: correctly calls the result of the election. It If you remove 0.75 percentage point of the vote of the winner, the result flips, and the winner no longer wins!

  • 40% or less: this market resolves to N/A. 3-way election.

Basically, if we are predicting someone wins by 80%, it has to win by a big margin.

I obviously understand that it's not how probabilities work. It's just a fun exercise.

I'll use the following market.

It basically means that if we predict with near certainty, the result needs to be really strong, where even substantial swings in voters preferences maintain the result.

I'll consider the final and official tallies for each state. I might decide to resolve it early if there's little to no uncertainty on the correct solution for this market.

Get
Ṁ1,000
and
S3.00
Sort by:

So,
If the favorite has >80%, this resolves yes iff the favorite ends up with a vote margin of 3+
if the favorite has > 60% but < 80%, this resolves yes iff the favorite ends up with a vote margin between 3 and 1.5,
if the favorite has > 40% but < 60%, this resolves yes iff the favorite ends up with a vote margin between 1.5 and 0.
Is this accurate?
Edit: multiplied vote margins by 2

Almost exact. The vote margins in each state. Therefore I'll move the vote margin in Kentucky and Pennsylvania by the appropriate margin and check whether the candidate still has 270 votes to win.

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