The article doesn't mention how they account for survivorship bias. Does anyone know?