MPT fails because we don't have an appropriate risk measure. In the real world there is no risk/reward trade-off over a wide rank of scenarios. The variance/covariance matrix is not stable. Asking people to come up with their risk tolerance without a sound way to measure risk is senseless. I think Mr. Mehrotra points this out well.

I agree that understanding variability in returns is useful for investors, but the profession has a lot of work to do in coming up with new ways to define risk. MPT is fatally flawed.