full replication vs stratified sampling.


It would be dependent on the case facts but if its around/over 1000 then it would be safe to go with optimization or stratisfied sampling.

Then to narrow it down between the two you would rely on the case facts. If client is concerned with tracking risk, then optimization would be better the better option. If they are concerned about transaction fees in replicating the benchmark then Stratisfied Sampling is preferred.

Optimization = Top-Down Approach

Stratisfied Sampliing = Bottom up Approach.

Hope this helps