Many excuses are made for continuing to use the old mean-variance method despite its obvious deficiencies, see for example this Note:
However, the reality is that it is simply due to the fact that fast and stable CVaR optimization is much harder to implement in practice than variance optimization.
This makes some people create rumors about CVaR optimization that are simply not true, for example, that “there are not enough observations” or that “CVaR does not work for high-dimensional portfolios”.
The Better Backtesting article coincidentally rejects the first excuse, while the Python code for this video rejects the second excuse.
As previously explained in the High-Dimensional CVaR Optimization article, CVaR actually becomes even more natural for large portfolios.
The video in this post goes through the Python code and summarizes all of the above. It is a part of the playlist that goes through the fortitudo.tech Python package (make sure to give this one a GitHub star if you want us to add more functionality).
Build a deep understanding
For a deep understanding of all the methods presented in the video above, it is recommended to study the Portfolio Construction and Risk Management book and its accompanying Python code.
The Applied Quantitative Investment Management course carefully goes through all this content. You can watch the first lecture for free, with the link available at the bottom of this post.
The remaining lectures are available for paid subscribers. Note that a paid subscription gives you access to all the paid content, including the expanding collection of exclusive case studies.
If you subscribe using your work email, consider expensing the subscription and potentially using the group subscription option to save 20% for you and your colleague.
Lecture 1: Intro and Python setup
This is the first lecture of the Applied Quantitative Investment Management course that goes through the Portfolio Construction and Risk Management book and its accompanying Python code.
