Open mjvakili opened 4 years ago
Thank you. You raise a good point, but we unfortunately don't have any plans to address the problem you pointed out. You can simply leave out the detoner for the problematic optimizers.
On Thu, Aug 27, 2020 at 7:20 AM Mohammadjavad Vakili < notifications@github.com> wrote:
A very nicely put-together library!
I have a question about whether you have any plan to modify the _get_optimal_portfolio in optimizer.py in cases where the covariance matrix is detoned. The detoned covariance matrix is singular. This is not an issue for clustering purposes (as in NCO and Hierarchical Risk Parity) but it is problematic in the allocation part where we have to take the inverse of the covariance matrix to calculate the weights. Particularly in NCO, this appears in calculating the intra cluster weights.
I was wondering if you have considered this or whether you have any plan to add this as an additional feature.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/enjine-com/mcos/issues/47, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAXTIXJ2PT2A56KUMF3OLB3SCZTTVANCNFSM4QNBQ3BQ .
--
Jin Choi, PhD
Founder/President
(226) 980-6902 - enjine.com
“Bring Your Quant Vision to Life”
A very nicely put-together library!
I have a question about whether you have any plan to modify the
_get_optimal_portfolio
inoptimizer.py
in cases where the covariance matrix is detoned. The detoned covariance matrix is singular. This is not an issue for clustering purposes (as in NCO and Hierarchical Risk Parity) but it is problematic in the allocation part where we have to take the inverse of the covariance matrix to calculate the weights. Particularly in NCO, this appears in calculating the intra cluster weights.I was wondering if you have considered this or whether you have any plan to add this as an additional feature.