An investor typically decides a portfolio objective and sticks with that objective for every rebalance date in the portfolio's life. Common characteristic portfolios that the investor chooses are:

  • minimum variance portfolio
  • maximum Sharpe (tangency) portfolio
  • inverse volatility portfolio
  • equally-weighted portfolio

Those familiar will know that the first two portfolios are solved with quadratic optimization while the last two are simple heuristics. I think that, because of changing market situations, no single one of these should be followed for the entirety of the investment life, through each rebalance date. Instead, rebalance horizon 1 might be appropriate for the first portfolio listed, horizon 2 should invest in the 3rd portfolio listed, horizon 3 should go back to the first portfolio listed, etc. In other words, the portfolio objective should constantly be reconsidered per rebalance date.

What dominant models are out there that, instead of focusing on stock selection within just one chosen characteristic portfolio, constructs a probability chart of which of the 4, or many, portfolio schemes should be followed at several rebalance dates as they arrive?

  • $\begingroup$ Have you considered a Black-Litterman approach that takes user input? $\endgroup$
    – user89135
    Jul 1, 2020 at 19:49
  • $\begingroup$ BL takes investor views about individual assets. completely irrelevant to question $\endgroup$
    – develarist
    Jul 1, 2020 at 19:51
  • $\begingroup$ You are asking for an investor input. How else does "the portfolio objective" get reconsidered? $\endgroup$
    – user89135
    Jul 1, 2020 at 20:06
  • $\begingroup$ look, its not the first time that someone has jumped up and said "black litterman" when asked about alternative portfolio models, and it doesn't make you look profound by just saying those two words in sequence. i think the decision to switch between different well-known portfolio schemes would be down to the ever-changing returns data per rebalance horizon, not investor views/user input. the returns would be the only 'user' input. your follow-up question shows that u don't even realize that portfolio optimizers solve an objective function to solve portfolio weight estimates. $\endgroup$
    – develarist
    Jul 1, 2020 at 20:14
  • 2
    $\begingroup$ I was just trying to help not "look profound". Good luck optimizing for maximal predicted returns with an automated regime switch based on past data that requires no user input $\endgroup$
    – user89135
    Jul 1, 2020 at 20:22


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.