Reproduce “Towards Calibrated Probabilistic Forecasts for Events of Interest via Outcome-Conditional Recalibration” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jakob Benjamin Wessel, Sam Allen
The authors introduce outcome-conditional recalibration, which adjusts forecast distributions within chosen outcome regions and separately corrects each region’s probability mass. In German electricity-price forecasts, it lowers negative-price calibration error from 0.0611 to 0.0051, while CRPS rises from 2.65 to 2.70. The result uses rolling recalibration; the guarantee applies in-sample to calibration data.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.