So they don’t have any tables describing their regression coefficients, but one of the params they use is lag-1 medals. For example, predicting 2020 medals usin

So they don’t have any tables describing their regression coefficients, but one of the params they use is lag-1 medals. For example, predicting 2020 medals using 2016 medals. It wouldn’t surprise me if the vast majority of the model’s predictive power came from that, as opposed to GDP per capita, whether the country is Muslim (yes this is actually a predictor too!)