1) Only if you change the relative proportions of the rest of the portfolio after including the asset. You can improve the risk-adjusted return, but the return of a portfolio is just the weighted average of the returns of its constituents.vineviz wrote: ↑Wed May 20, 2020 4:19 pmYep, pretty much. With high enough volatility, low enough correlation, and periodic rebalancing you can increase the return and lower the risk of a portfolio by including an asset with an expected return of zero.
The annual return of a portfolio of 80% VBINX, 20% Whatever, equals 80% of the return of the VBINX + 20% of the return of Whatever. If Whatever has lower return than VBINX, the return of the new portfolio will be lower than the return of VBINX.
There's nothing profound about this and it should go without saying, but the rhetoric used by some advocates doesn't always make this clear.
If adding Whatever reduces the volatility of the whole portfolio enough, you can take the improvement in the form of higher return rather than lower volatility--but you need to change something. The most obvious change is to use leverage on the whole portfolio and lever it up until it has the same volatility as before, and it will have higher return. Depending on what is inside VBINX, you may also be able to do it by modifying the composition of VBINX. For example, VBINX is actually a 60/40 stock/bond portfolio, and you might be able to overcome the drag of Whatever by raising the stock allocation.
2) And only if "low enough" means "negative." It is theoretically possible to add a zero-return asset to a portfolio, adjust the proportions of the rest of the portfolio, and construct a new portfolio with higher return, but no higher volatility, than the original portfolio. This may require the use of leverage or short positions. It definitely requires negative correlation--not just low correlation, but negative correlation. In other words, when you say "low enough," you implicitly mean "negative."
3) I don't think robust negative correlation, over more than... let's say five years... has ever been seen between any two stock categories. Please educate me if I'm wrong about that.