Ernie Chan's Kelly for the portfolio uses a fraction of 1 whole

AI & Machine learning
1

Ernie Chan's Kelly for the portfolio uses a fraction of 1 whole. 

 

http://epchan.blogspot.com/2014/08/kelly-vs-markowitz-portfolio.html  

But in your course: Quantitative Portfolio Management  

Section: 4  

Kelly Criterion  

 

Your Kelly criterion for the portfolio adds up to 1 whole or 100% How come your version of the kelly portfolio weight total is equal to 100% ?  

  

How can I calculate the Kelly /F on the portfolio to reflect the overall maximum exposure? Using the library provided? 

 

Thanks again in advance for all the help.  

Namaste. 

2

Hi Jane,



In the Quantitative Portfolio Management course, the total weight of the portfolio is fully invested, with no cash left unallocated. This is a common approach in portfolio management, as it simplifies the implementation of the strategy and makes it easier to track performance.



To calculate Kelly /F you have to define the expected returns and covariance matrix for the assets, as well as a maximum exposure constraint, and use cvxpy to calculate the optimal portfolio weights that maximize the expected return subject to the constraints.



Hope this helps!



Thanks,

Rushda Ansari

3

Can you show an example, and post the code using cvxpy  ? This is the a major focus point of using the kelly criterion. F is often times used as a measure of exposure or a limit for max exposure. 





 

4

Hey Jane,



You'll find example of using cvxpy here - Examples -



These examples would help you write your own code.



Thanks,

Rushda Ansari

5

You need to put more effort into providing a decent service for clients. Why would you not cover an example when thats how Dr E Chan uses it?

6

Dear Jane, this forum is NOT a client service platform. And it is you who need to put in more effort here. This is a community where your fellow quants help each other out. While the quantra team do their best to answer questions directly related to a course, if you need anything above and beyond, you need to put in your own effort to make it worth their (or any other user's) while. On this forum, you should not expect answers where 1) your questions are badly framed/ unclear/ irrelevant or 2) you are looking for shortcuts and asking for readymade answers without putting in minimal effort.



I gather you have enough understanding to appreaciate that If a course tries to cover all the possible ways a particular topic is treated in the quant world, it will be a reference document, not a course. I leave it to your understanding to figure out what to do, if you want more help on this topic (hint: try doing it yourself first).