The Will Rogers Phenomenon

Named after the comedian's joke that when Okies left Oklahoma for California during the Dust Bowl, they raised the average intelligence in both states — the Will Rogers phenomenon describes a real statistical effect: moving a borderline member of a low group into a high group can raise the average of both groups, even though every individual's own value is unchanged.

It happens whenever the item being moved is better than the average of the group it's leaving but worse than the average of the group it's joining. Removing it lifts the group left behind (it was dragging the average down); adding it also lifts the group it joins less than you'd fear, or even helps, if it still beats that group's worst members. It shows up constantly in finance: reclassifying a borderline holding from one risk bucket, benchmark, or credit rating to another can improve the reported performance of both the old and new bucket without a single position actually changing.

Worked example. A fund's "core" holdings average a 6% return and its "opportunistic" bucket averages 2%. One position earning 4% gets reclassified from core to opportunistic (it was dragging core down, and it's still above opportunistic's average). Core's average rises above 6% with that laggard removed, and opportunistic's average also rises above 2% with a 4%-earner added — both buckets look improved, purely from where the line was drawn.