Are interceptions random events?

They are difficult to predict, that’s for sure

Jason and I have written about interception rates before. Over at Footballguys.com, I concluded that interception rates were almost entirely random from season to season. Jason concluded that among the most popular stats, interception rate was by far the least consistent when a quarterback changed teams, indicating that either interceptions are largely random and/or aren’t mostly the fault of the quarterback. He reached essentially the same conclusions when he analyzed what happened when teams changed quarterbacks.

 

I looked at every 16-game season in NFL history and noted every QB who threw at least 150 passes in his team’s first 8 games and at least 150 passes in his team’s last 8 games. There were 533 such quarterbacks. I then calculated each QB’s interception rate, relative to league average, in the first and second halves of the season. So how closely correlated were INT rates in the first and second halves of the season? Not very. The correlation coefficient was just 0.12, which shows a very small relationship between the two numbers. The R^2 was just 0.01, which means that (in loose mathematical terms) only 1% of a QB’s interception rate in the second half of the season can be predicted by his interception rate in the first half of the season. Or in other words, they’re almost entirely random.