I have in my notes a sentence regarding stratified sampling which says: - can be used where an index has a few components or a few large components Now I would have thought that sentence is incorrect - I would have thought that stratified sampling wouldn’t be very effective where an index has only a few components or a few large components?
stratified sampling (aka cell matching) would be appropriate where: - no of securities in the index > 1000 - a few securities might be iliquid
thanks L3aspirant Gone back through the text and it in fact does state that stratified sampling can be used where you have an index with few securities
agree. so when no of securities are fewer, we have the option of full replication (FR) or stratified sampling (SS) but when no of securities in the index are large, we are better of with SS bcoz FR will be more expensive to administer and more difficult to construct.