Conceptually, what are the drawbacks / unforeseen risks of running a portfolio whose weight are derived from what would have maximised the sharpe ratio over the previous time period (last 30 days) ?

  • $\begingroup$ This is quite broad but does this help: quant.stackexchange.com/a/43695/848 ? $\endgroup$
    – Bob Jansen
    Jan 29 '19 at 13:07
  • 1
    $\begingroup$ How well can you estimate expected returns and covariance using 30 days of data? $\endgroup$ Jan 29 '19 at 14:18
  • $\begingroup$ Good question Matthew Gunn. I want my statistics to respond to current observed market moves. I arbitrarily picked 30 days. I am afraid that if I go for a longer data set, what I will capture is event that happened in the past and are less likely to influence current market condition. What should I use instead, knowing I want to keep things simple for now ? $\endgroup$ Jan 29 '19 at 14:25
  • $\begingroup$ Expected returns estimated as actual returns over the last 30 days are likely to be almost worthless as a predictor. No better than random numbers. You may want to think of other ways of coming up with expected returns. $\endgroup$
    – Alex C
    Jan 29 '19 at 16:14
  • 1
    $\begingroup$ Here is a paper in which short lookback windows (up to 12 months) are used to construct tangency (max Sharpe ratio) portfolios. In essence, this is a momentum-based investment strategy. The backtest results presented in the paper are fairly encouraging. papers.ssrn.com/sol3/papers.cfm?abstract_id=2606884 $\endgroup$
    – MGL
    Jan 29 '19 at 16:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.