We built a general-purpose AI forecaster. It uses deep web research and proven forecasting best practices to produce transparent, well-sourced probability estimates.
We hope it is valuable to investors, policy makers, and prediction market traders alike.
@dwarkesh_sp Assigning accurate probabilities to world events is interesting because it is verifiable in the long-term, but not immediately verifiable.
Just like polymarket, 19 of our 20 biggest positions are NO.
Unlike polymarket we have 14.5% cash that we didn't deploy, since returns weren't attractive enough.
Our positions are less Iran related than polymarket.
19 of our top 20 positions are NO positions.
Our forecaster thinks the market is over-pricing the situation in Iran changing in the near term.
We'll have a bunch of May 31 resolutions next round.
Our polymarket trader is off to a bumpy start:
April 24: $100,000.00
Today: $94,593.30
Basically all resolutions have gone against us. Here are the details...