How to Improve the Performance of Portfolio Managers?

The main idea of combining universal enforcement learning with the enforcement learning is very general and so you can use it for any asset classes. The only input which is provided externally which we do not supply is the model for expected three choice for this asset class. So if you have a reasonably decent model that you use for other applications maybe then you can also use it here. And how is this supposed to improve the performance of individual managers and in fact how do you define improvement in this context? Well in the paper we showed improvement in terms of backtest rightSo we show that if portfolio managers that we kind of learned from collective intelligence of all fund managers in a certain group right.