What Is Survivorship Bias

Survivorship bias in five lines

Survivorship bias is judging a group by the members that survived, when the rest are simply not in the room. It hides in the rules that get published, the funds that are still listed, the traders who are still posting. It fools because the survivors all look good, which is exactly how they became the survivors. The cure is to count the ones that started rather than the ones that finished, and to ask where the rest went. In the example, 276 buy-and-sell hour-pairs were tried on the euro for one week of August 2026 and 243 of them made money; 21 of those 243 made money the week after; and only the 243 would ever have been talked about.

The planes that came back

Picture a plane that came back, with the holes it came back with, in the wings and the tail. The first thought is to armour the wings and the tail, because that is where the holes are. The right thought is the opposite: those holes mark where a plane can be hit and still return, and the planes hit in the engine or the cockpit are not on the runway to be counted. Every survivor you can examine has already passed the test that removed the others, and what it shows you is what passing looks like, not what the test was.

This paper is locked