I try understand the difference between prediction and confidence intervals. Say we want to predict the winning time of the 100 m race in the Olympics 2020. We have fitted a linear regression model of times from 1900 to 2016 and we want an associated 95 % confidence interval of our prediction time. Would we use a prediction or confidence interval? Should we use the prediction interval since we predict the value of a random variable and thus have an extra uncertainty?