All together now: the most trustworthy covid-19 model is an ensemble

Earlier this spring, a paper studying covid forecasting appeared on the medRxiv preprint server with an authors' list running 256 names long. At the end of the list was Nicholas Reich, a biostatistician and infectious-disease researcher at the University of Massachusetts, Amherst. The paper reported results of a massive modeling project that Reich has co-led, with his colleague Evan Ray, since the early days of the pandemic. The project began with their attempts to compare various models online making short-term forecasts about covid-19 trajectories, looking one to four weeks ahead, for infection rates, hospitalizations, and deaths. All used varying data sources and methods and produced vastly divergent forecasts.