Mean-Gini portfolio optimisation (Excel)
HTML-код
- Опубликовано: 10 июл 2024
- Today we are going to discuss a very unique and not so well-known portfolio management framework: the mean-Gini (MG) optimisation theory developed by Shalit and Yitzhaki in 1984. It utilises the Gini coefficient, a well-known inequality measure from economics, to measure the risk of the portfolio empirical return distribution. We will implement the mean-Gini framework in Excel, discuss its applicability, assumptions, and limitations.
Don't forget to subscribe to NEDL and give this video a thumbs up for more videos in Investment!
Please consider supporting NEDL on Patreon: / nedleducation
You can find the spreadsheets for this video and some additional materials here: drive.google.com/drive/folders/1sP40IW0p0w5IETCgo464uhDFfdyR6rh7
Please consider supporting NEDL on Patreon: www.patreon.com/NEDLeducation
Hi, thankyou for your video, it is a very interesting material explaining a new method of portfolio optimization (new for me). But I have a question on the Gini coefficient formula. When I look up to the original journal by Shalit & Yitzhaki (1984) the Gini coefficient formula is only covariance between return and its rank. Why did you divide it by mu times n? Is there any reason behind it?
Thank you, this is very helpful! Do you have a video on how to construct a mean semi variance efficient frontier. O learnt that the approach is not so exactly with the MV approach due to the endogeneity problem encountered in the ES approach
Hi, and glad you liked the video! In principle, it is, but there is no easy closed-form solution as there is for mean-variance (not sure whether mean-semivariance would be harder than mean-Gini though). I do touch upon the closed-form solution for mean-variance here (for Excel): ruclips.net/video/fjEkkVwRl2A/видео.html and here (for Python): ruclips.net/video/ymM0xJUgGLo/видео.html. However, the mean-semivariance frontier can always be calculated using numerical optimisation, might do a Python video on that at some point!
Hey Savva, mentioning optimization... ehemm.. sorry, the British way, optimisation :P, it would be helpful to cover optimization methodology and processes that are performed behind the scenes for Excel's GRG nonlinear and Evolutionary methods within Solver. While I have a feel for when to do what, I don't entirely understand what happens behind the scenes with each method or input option.
Hi Stephen, and thanks for the question! Long story short, GRG is basically gradient descent - the algorithm looks in which direction locally the function grows (or falls) the fastest, moves in such a direction a little bit, and reiterates until convergence. Evolutionary method is more brute-force, hence why it requires restrictions on all variables to achieve the result in finite computational time. I might do a video on manual gradient descent implementation at some point in the future. Hope this helps!
Another nice video. Can you produce a video on the Hawkes Process?
Hi, and glad you liked the video! To be honest, this is not something I have looked in particular depth into, but someday I might research this concept more rigorously and make a video out of it, who knows :)
@@NEDLeducation I don't know if I can add a RUclips link in this comment, so I'll just describe how to find the video. In RUclips, search for "Mathieu Rosenbaum". The RUclips channel is Conférence GP60. The video is in English after time 3:15. The Hawkes Process discussion starts at time 12:19, describing a high frequency trading model based on the Hawkes Process. The result is a Rough Heston Model (time 19:12). What's important is that the Hawkes Process seems to do a reasonable job at describing the micro-structure of a market. If that's true, then it is worthwhile to explore the process in more detail so that the average viewer can get a feel for what's going on in the trading model.
Thanks,
Bill
Just in case a link can be posted, here's the link to the above mentioned video.
ruclips.net/video/1ztrf3byQe8/видео.html
Great work. I wonder if you would consider repeating the work in python ?
Hi, and many thanks for supporting the channel! Yes, a video on portfolio optimisation techniques in Python is in my nearest plans, stay tuned :)
@@NEDLeducation Awesome! Thank you for your work
Can you tell me how construct sized based portfolio by ranking all the firm in the beginning of each year according to their market capitalization and then divide them into five equal quintile group. kindly guide me regarding this it help me alot i shall be very thankful to you for this favour.
Hi, and thanks for the question! I presume you mean Fama-French factor portfolio sorts? I am planning to do a video on these at some point in the future!
@@NEDLeducation yes Thanks alot