[Retric]

I strongly disagree that using such methods to calculate a minimum variance optimization is a reasonable benchmark for portfolio performance. The issue is limited historic data is a very poor fit for future risks.

Looking at gold over the last say 2,000 years years shows a very bumpy ride with large long term negative returns. Stock data doesn’t have anything close to that kind of history, but looking at various historic stock markets again shows a lot more variety than simply reviewing a winner like the US stock market.

Essentially, with bad enough assumptions or data any calculation is meaningless.

[incognition]

I agree with this analysis. Anybody going off on Sharpe ratios has to extend that analysis to long tailed distributions, which venture investing certainly is. For example, if you do actually hit the Uber of companies, one early investor in Uber put in $5000 and reaped 20 million.