One Wonders, from Victor Niederhoffer
One wonders if there is a regression bias effect so that all numbers, indeed all events that are expected to be very much better than the last occurrence tend to disappoint. There are random factors as well as recurring "skill" factors in all such events. The random factors bring the actual number down towards the mean while the skill factors tend to bring it up. Assuming they are both equal in a number like employment, the expected realization is half way between the previous and the predicted.