Equity Portfolio Mgmt-Optimization vs stratified sampling

Hi, Equity Portfolio Mgmt- (i)Full replication done whenver securities are liquid. But if it is illiquid-would optimization or stratified sampling-be preferred-is there any basis to chose between the two.

i don’t think that a preference is given in the material and I’m almost sure it isn’t in the secret sauce.

I thought I remembered reading that optimization was better, IF it could be implemented correctly. It takes correlations of assets into consideration and will result in less issues need to replicated the risk factors for the index. Bankin’ is on top of his stuff though so I could be wrong here though.

I don’t think there is a real world example of full replication b/c it is costly when manager can simply buy futures or invest in an index directly. If manager wants to achieve alpha but keep tracking error and cost low, then stratified sampling may be preferred.

mwvt9 Wrote: ------------------------------------------------------- > I thought I remembered reading that optimization > was better, IF it could be implemented correctly. > It takes correlations of assets into consideration > and will result in less issues need to replicated > the risk factors for the index. > > Bankin’ is on top of his stuff though so I could > be wrong here though. That could be, I don’t have the study notes or the curriculum at the office and it could have been left out of SS, but I just checked in the Secret Sauce (p. 156) and it doesn’t say anything about it. It does say “A replication fund will underperform the index to a greater extent when the underlying stocks are illiquid…” But liquidity isn’t mentioned anywhere in the description, advantages, or disadvantages of stratified sampling or optimization.

I’m not sure which one is preferred. Optimization has the benefit of having fewer assets and smaller tracking error. However, the optimized portfolio needs to be rebalanced. Stratisfied one has larger tracking error but doesn’t need to be rebalanced.

I think Stratified Sampling is easier to implement than optimization. Optimization is quant heavy stuff and not every firm would have the resources to do it. Stratified sampling is easier.

as noted above, optimization requires frequent rebalancing and therefore if the portfolio is illiquid, it wouldn’t be a good choice due to high transaction cost. You’d want to stratify…

I remember optimization has smaller track error than stratified sampling, because stratified sampling ignores the correlation and other links between individual assets. There may be no “absolutely better” and “absolutely worse”, but I want to remember the characteristics of each method.

Thanks guys. I think-such that material doesnt touch upon the relative advantages and optimization has higher transaction costs-than stratified sampling, stratified sampling should be preferred, when the stocks are illiquid