2016 PM Mock #11 answer says that stratified sampling has lower transactions costs than optimization. I can’t find an explanation for this in the reading. Anybody know why?
Use Russell 2000 as your example, with some relatively illiquid securities.
#ThinkAboutIt
IMHO, in extreme case, the results of optimization is that you should buy many small stocks which have illiquidity problems causing transaction costs up and making it impossible to excute orders. But stratified sampling may help the results be less tilting to expensive stocks in terms of transaction costs(either explict or implict) by slicing whole samples based on some criteria such as size, ratios etc. // If it is not right please let me right thanks.
i was confused by this question too. Thought that optimization should generate less trades and rebalancing b/c of better matching to the risk factors and lower tracking error.
Optimization needs constant rebalancing to keep the risk factors matched, hence the higher transaction costs.
Stratified sampling is basically hand-picking.
It’s a poor man’s index replication, cheapest and least effective.
in the real world it would. you would not write an optimization program that didn’t consider all costs (direct & indirect) and style exposures.
the author is trying to argue that picking stocks randomly will pickup about 90% of the systematic risk of an index. sampling manages the style exposures…