Tracking error about riskfolio-lib

[dcajasn]

Hi Tom, the formula that I use for tracking error is to calculate the difference between the index returns and portfolio returns and then calculate the sample standard deviation of this difference. I use this approach that is equivalent to the ex-ante because it can be used when you don't have the composition of assets of an index. Also, the tracking error is a constraint, so the resulting portfolio must have a tracking error lower or equal than your constraint. Another problem that you must consider is that the tracking error constraint must be in the same frequency than your assets returns. If you have a problem, you can send me your notebook to check where is your problem.

[TomLiu518]

Hi Dany, thanks. I guess my question now is, can you customize constraint and objective functions? for example if i want use a different method to calculate tracking error as a constraint?

[dcajasn]

Hi Tom, about your question you can't customize objective functions. The reason for this is that Riskfolio-Lib is based on cvxpy so it can only solve disciplined convex problems. All combinations of objective functions and constraints in Riskfolio-Lib are disciplined convex problems (you can check the papers in documentation), so they can be solved by cvxpy. On the other hand, if you want to limit the max value of a risk measure while optimize another is posible (I will make an example soon), for example you can maximize sharpe ratio while you put a constraint on CVaR. Finally, If you want a new objective function or constraint, check if they are convex and send me the formula or paper is based to try to include in Riskfolio-Lib.