Problem
1. You read a journal article that describes an investment strategy. In a backtest, it achieves an annualized Sharpe ratio in excess of 2, with a confidence level of 95%. Using their dataset, you are able to reproduce their result in an independent backtest. Why is this discovery likely to be false?
2. Investment advisors are plagued with conflicts of interest while making decisions on behalf of their investors.
(a) ML algorithms can manage investments without conflict of interests. Why?
(b) Suppose that an ML algorithm makes a decision that leads to a loss. The algorithm did what it was programmed to do, and the investor agreed to the terms of the program, as verified by forensic examination of the computer logs. In what sense is this situation better for the investor, compared to a loss caused by a discretionary PM's poor judgment? What is the investor's recourse in each instance?
(c) Would it make sense for financial advisors to benchmark their decisions against the decisions made by such neutral agents?