Will the mean squared error of the predicted to the actual Brier score for the questions be better on logistic fit than the exponential fit, for all resolved Metaculus binary questions?
Created by niplav on 2022-09-30; known on 2024-01-01
- niplav estimated 60% on 2022-09-30
- niplav said “Prediction on replicating results from https://niplav.github.io/range_and_forecasting_accuracy.html with full Metaculus data. Accuracy between forecasts. #metaculus” on 2022-09-30
- niplav said “No, their mean squared errors are *equally good at predicting the data (~0.02445 in both cases).” on 2022-10-03