PM struggles with a large number of strategies. PM is such an important piece of functionality it seems a pity not being able to load it with 1000s of strategies. I think the following draft use case would add a lot of value and make it much easier to work with a large number of strategies:
1. Load PM with large number of strategies, say 10000.
2. Start PM.
3. Each time PM finds a portfolio of a given size, the component strategies are deleted from the data bank. This would continue until all the strategies are deleted from the databank or a given condition is met.
4. All the portfolios are then split to singles strategies, go to 2.
5. after the number of strategies reach a manageable size (computer resource wise, say 200 strategies). The standard brute force/genetic search would continue as normal.
Status changed from New to Fixed
We will improve PM feature to work with large number of strategies in the next version.
I tested the suggested process earlier today. Quickly generating a portfolio of ten strategies, deleting those ten strategies then repeating the process tends to weed out the highly correlated strategies. The 10 strategy portfolios strategies are then added back to the databank and then larger size portfolios are much quicker to find.