I hope this does not close because it is related to algorithms that I could not figure out (it is also quite a long time because I am so confused about how this is done). Basically, many years ago I worked in a mutual fund, and we used various tools to choose portfolio optimization, as well as hedge existing ones. We will take these results and make our own changes, and then sell them to customers. After reducing my company, I decided to try (to create software and enable my settings), but I don’t know how combinations are created for software.
After 6 months of trying, I agree that my approach is impossible. I tried to use combination algorithms, for example, from Knut’s book, as well as a combination bit, to try and find all possible portfolios (I limited them to 30 shares) on the NYSE (5000+ shares). But according to everyone I spoke with, it would take me billions of years to get only one result (for me, I stopped it on the GPU after 2 days of direct processing).
So what am I missing? We enter into our risk tolerance and view of the market (expectations of stock market growth, expectations of inflation, expectations of fund expectations, etc.), and this will give us an ideal portfolio (theoretically ..) in a few seconds / minutes. With thousands of possibilities and quadrillion possible combinations of weights, how can they quickly calculate results (or even in general)? As a system administrator, I know that we upload a file every day (less than 100 mb and upload to the mssql database, possibly only market data), and this does not seem like we had all the features. Using my approach above, I would get 5 gigs of the file per minute of execution of my version of the Knuth keyboard shortcut), and the applications were working offline (so he must have done it locally on a desktop / laptop,and not on a massive supercomputer somewhere and took a minute or two. minutes was the longest for the global fund, which includes all shares in the world). This is so confusing because their work required correlation of the entire fund (I don’t think that they simply sent the top shares, which they tentatively calculated, because everyone had different results). Therefore, if I wanted the 30 fund to give me 2% of the profit and have a negative correlation with the market and be hedged by 60%, how could the software generate this portfolio of billions of opportunities so quickly? Notice, I’m not asking about math or the financial part, I’m asking how it could generate 30 shares from the whole market, which gave 2% of the profit,when for this he needed to know the profitability of all 30 portfolios of stocks (only this would make him work for billions of years, since other restrictions make it more difficult).
, ? , Knuth, , , , , . ?