Based on teachings by Kora (@paststat on twitter) via his watanabe you tube show, had made a function to run create the optimal portfolio weights for a given portfolio. The post on twitter had gained a good response, since I’d made the code open source.
However, later I realized that the tool might not be used, since most people are not R programming savvy. To improve upon this condition, decided to run the analysis daily on Index data of NSE to provide the optimal index weights based on last 20 trading sessions (one month calender time). Cannot be called a major Alpha Leak, but will provide some utility to those who are interested. The code will update automatically everyday to change portfolio weights, if necessary.
Those who wish to follow it, can bookmark the page by Ctrl+D on their browser.
Best Risk adjusted Weights for last 20 days
Best weights for Max Returns in last 20 days
I ran the same 20 day mean variance portfolio as a rolling window since 2012. Notice that the changes are sometimes abrupt, sometimes slow. Still need to validate if these changes have any bearing on forward returns. That’s for some other time. For now, enjoy the view (like I said, not a real alpha leak, but definitely interesting) – Best viewed on computer screen.
Right now, for quite some time, its 100% in Autos (Disc. that’s where I’m positioned too – intuitively, later got the pleasant surprise seeing this result).
Those who wish to check out Kora’s lecture on Mean Variance Portfolio optimization on Excel, refer to the video below.