First time that I have encountered p-values <<10⁻³⁰ irl

Image description: Plot with four lines. Two are correlation coefficients, of Metaculus truncated correlations and PredictionBook truncated correlations. The Metaculus correlations are close to zero in the first ~15 months, then dip into negative correlations (around -0.2) until month ~35, then rise to positive correlations (around 0.15) until month 40, and then start oscillating wildly (shortly afterwards the data for Metaculus correlations ends).

Follow

Image description (cont): The PredictionBook correlations also start around 0, then rise slowly to ~0.05 at 60 months, at which point they start oscillating around 0 with larger and larger amplitudes until month 120. The p-values for Metaculus start out around 10⁻⁶, then jump around between 10⁻⁵ to 10⁻² in the first 15 months, and then dip down in the range of 10⁻²⁵ to 10⁻³⁵, and then recover back to 10⁻⁵ to 10⁻² until the end of the Metaculus dataset.

Image description (cont): The PredictionBook p-values start out at around 10⁻⁵, where they stay until 25 months, then rise to 10⁻¹ to 10⁻² until month 40, then drop down back to 10⁻⁵ until month 65, and then finally rise back up to 10⁻² to 10⁻³ month where they stay until the dataset ends.

Sign in to participate in the conversation
Mastodon

a Schelling point for those who seek one