He has the advantage of being able to consult his forecast, but sometimes he has hot takes that contradict it. Will those takes be vindicated?
In 2024, Nate wrote two weeks before the election in the New York Times that his gut leaned Trump — but by Election Day his instinct switched to Harris, and he said he'd maybe put her odds at 52% instead of his forecast's 50.015%. It was a close situation, but I resolved the corresponding market NO.
In 2022, Nate wrote an article called "Why I'm Telling My Friends That The Senate Is A Toss-Up" (https://fivethirtyeight.com/features/the-senate-is-a-toss-up/ ) when his forecast showed Democrats leading, but his forecast turned out more correct.
In 2020, Nate sent this tweet after the election, which I think captures the subject of this market pretty well:
I will do my best to resolve in good faith, and won't bet after initializing the probability. It would have resolved NO for the 2020, 2022, and 2024 elections since his forecast was more accurate than his own personal opinions.
If Silver does not release any kind of midterm election forecast, this will resolve N/A.
General policy for my markets: In the rare event of a conflict between my resolution criteria and the agreed-upon common-sense spirit of the market, I may resolve it according to the market's spirit or N/A, probably after discussion.
Update 2025-09-15 (PST) (AI summary of creator comment): - If there's a true tie between Nate Silver's personal predictions and his forecast, the market may resolve 50% (PARTIAL).
The creator will try to break near-ties by examining details to determine whether Nate or his forecast performed better; only if it's genuinely indistinguishable will it be treated as a tie.
Update 2026-11-05 (PST) (AI summary of creator comment): The creator has announced they plan to resolve the market to NO.
1,000🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ333 | |
| 2 | Ṁ11 | |
| 3 | Ṁ1 |
From this pre-election day post:
Overall, though, I see FLIPR’s win probabilities as about right, just more asymmetric than FLIPR has them. If I were placing my bets, I might wager some chips on the “Dems slightly underperform polls but win the Senate anyway” spot on the felt and some on “Republicans narrowly hold the Senate,” but also a few on an epic Democratic blowout.
I feel like this is evidence for YES, because he is expressing higher variance than the model — putting a few chips on an epic blowout, which I think implies he thought it was more likely than FLIPR. However, Claude disagrees and thinks this is evidence for NO because he seems to think the more likely miss is Democrats underperforming.
From this post today:
I am interested in the differences, however. I discussed the Senate races where prediction markets and FLIPR disagreed in a post on Oct. 25. The funny thing was, in several of the races like South Carolina, FLIPR was more accurate, but my “gut” was on the wrong side of FLIPR! Prediction markets do deserve some credit relative to FLIPR on Mississippi, though, where Scott Colom is probably going to win in a major upset.
Seems like a pretty relevant statement. South Carolina was a race to trust FLIPR on!
I gave my Claude that post and it made a chart of Nate's gut vs FLIPR for the other races. This seems like a pretty clear NO to me.
Overall, I'm going to resolve NO.
@Conflux, what if he ties his forecast. e.g. he says that his gut says the democrats will do better in the house and senate than his model, and they do better in the house, but worse in the senate.
Also, this has 2024 election tags, shouldn't that be 2026?
@DanielTilkin Thanks for the tag correction! I fixed those.
If he truly ties his forecast I would be willing to resolve 50%. But I will be reluctant, I will be looking to go into the details of whether he or his forecast was better.