Thanks for the links, but... The vendor doesn't explain the algorithm his optimizer uses -- unless I missed something. That's not of much value (although, hey, it's free). Have you actually used or reversed-engineered this tool?
This tool's(http://quicark.com/OptimalPortfolio.aspx) principle is: if prices of securities happened in the past, it will happen in the future in a similar way. From this point of view, we can value the risk as the value fluctuation diverse from the mean route of the portfolio. The objective of the optimization is rrr * rate â risk. Where rrr is the return risk ratio, return is the expected weighted rate of return, and risk is the risk of the portfolio.
I am not sure if the objective is correct. Regardless, since this is a higly non-linear function, what optimization algorithm is used and how do you know that it does not reach a local optimum point?
I agree that the object is non-linear. At least the rate of return part is linear. I cannot expose too much about the detail of the algorithm. You can change the rrr and play with it to see if it is a local optimal point.
If you add a linear function to a non-linear one you get non-linear. How do you optimize that? What initial feasible point do you use? What type of algorithm do you use? I am not asking you to give details. just the general strategy. Otherwise, what you are doing is just an ad-hoc solution that is far from optimal, maybe even non-optimal.
To my taste, then it's worth no further discussion. We live in the FOSS era. There are plenty of open-source, fully documented tools available. Life is too short to spend time fiddling with mysterious blackboxes.
Well said and to the point. No trader should waste one minute discussing black-boxes. They always hide skeletons inside.