Full replication? Optimization? Or stratified sampling?
I don’t know how many stocks Russell 2000 has but if it is 2000 (http://www.russell.com/indexes/data/fact_sheets/us/russell_2000_index.asp), you have about $5000 per stock if they were equal-weighted, and they are not. So you’ll probably have $100 to buy some stocks, barely enough. The question is, is that enough for full replication?
On a relevant note, just a quick question. If the goal is to minimize TE:
if a portfolio was reconstituted annually, but changed to quarterly (more frequent). Would that help in minimizing TE? I would say yes, but not 100% sure.
For me, it’s stratified sampling. Can’t possibly do full rep at that amount. Get your sectors roughly in line and that’s the best you can do. Unless you wanna put it all on red.