The Signal and the Noise: Why So Many Predictions Fail-but Some Don't

by Nate Silver

The Pareto Principle of Prediction implies that the worst forecasters—those who aren’t getting even the first 20 percent right—are much worse than the best forecasters are good.