Okay, let's be very precise.
Modeling free throw percentage as a random process with some intrinsic mean and variance, it appears likely that the 28/51 Ace is currently shooting from the line does not reflect his the actual mean of his free throw percentage. Instead, given that there is a larger sample of him making free throws at a much higher percentage before college, it makes more sense to estimate the parameter of his mean free throw percentage as closer to 80% than 55%. If he continues to shoot poorly from the line in college, then the estimate for the parameter of his mean free throw percentage will have to be revised downwards.
This modeling process is used more predicatively than descriptively as there are essentially no truly random events in basketball and with every single variable modeled, every outcome could be predicted perfectly. However, this is currently impossible and we can assume that the factors that cause made or missed free throws follow a random process centered around a hypothetical mean value that varies from player to player.
Or are you now going to get mad about me using big words after you got mad at me for using words colloquially?