Yes, there are many rules of thumb but almost all are poor. The big problem is the efficient frontier theory is pretty good but as you found is extremely sensitive to the inputs. Since the (medium-term) future often looks very different from the past in finance the inputs are often off and produce extreme portfolios. This is one of the hardest questions in finance.
There have been some efforts to take this uncertainty into account, Bayesian portfolio optimization, mini-maxing, risk-parity are some proposed solutions but all have similar sensitivity issues. Hedge funds sometimes use interesting, more complicated methods but generally have very specialized portfolios and these methods don't apply generally.
Honestly, even at the highest levels of finance practitioners often start with a tried and true portfolio like the 60/40 with a risk profile that they are comfortable with. Then they add different amounts of additional investments they think will add value add see how it would have changed their portfolio over different past periods and different stress scenarios until they are happy with the risk and return that these investments would add.
For example, with my own portfolio, I looked at gold and found that the returns are extremely uncertain and theoretically dubious and found it had no place in my portfolio. However, I found REITs had a rough period recently but reasonable expected returns with similar risk profile to my equity exposure. So after a number of historical checks, REITs found a small allocation in my overall portfolio replacing some of the equity component.
The Vanguard link above is great reading to learn more.