Will Nate Silver’s model correctly predict the election?
Basic
174
45k
Nov 6
65%
chance

Resolves YES if the winner of the 2024 presidential election is the candidate predicted most likely to win by Nate Silver’s model (released today, here) on Election Day. NO otherwise.

Get Ṁ1,000 play money
Sort by:

Slightly better than 50/50 is unfortunately the best humanity has to offer in terms of prediction models.

I mean… Nate had Trump at 30% in 2016 which would have resolved NO by this criteria even though 538 gave Trump a better chance than basically anyone.

yeah I feel like this bet is exactly the kind of interpretation of his model that Nate tries to hard to avoid 🙃

This is just the probability Nate's favored candidate wins according to the best calibrated model right before election. If you believe Nate's model is perfectly calibrated, this should be identical to his predicted win probability. If you believe an alternate forecast/market is better, you predict what that forecast will be on the final day, assuming Nate favors the same candidate. In 2012, markets had Obama at 73%, 2016 had Clinton at ~80%, 2020 had Biden at ~60%. I think this election is closer than 2012 and 2020, so 60% seems fair at the moment

this election is NOT closer than 2012 and 2020

I take this back this election just got a lot closer

And in the unlikely event they’re tied in his model?

that's a roughly 1 in 40000 chance, but in that case it should resolve to 50%.

If we include statistical ties (49.5% to 50.5%) it's more probable. Perhaps 50% if Nate Silver says its tied (similar to 538 is doing currently)

In the unlikely event there is a statistical tie, that would mean that the model can’t predict who will win.

I think in this case the market should just resolve NO.

Counterpoints?

Refining my answer to the question of how this resolves if there is a “statistics tie”.

First, note that Silver rejects this term:

A statistical tie would traditionally be based on a margin of error, which is a measure of uncertainty. It’s traditionally accounted for with variance, a measure of randomness within a sample.

Silver’s method is instead to run enough simulations to control for uncertainty by factoring it in empirically, e.g. as the probability of a specific outcome, like Biden or Harris or Trump winning given the variability of all the possible factors along that, such as the economy, the Electoral College, global conflicts, etc.

Thus it isn’t meaningful to worry about a statistical tie in Silver’s model.

In the case of a literal tie, where Trump and Harris have the same probability of winning down to all significant decimal places, this market would resolve NO.

Predicted as of when?

Market close. Nov 5. Election Day.