"Risk depends on the resolution". Great Comment. I invest in real estate LPs. They run 3-6 years and have no intermediate liquidity, and thus no perceived volatility. I review the financials, and operating reports, but there is no mark to market.
On the other extreme 3x leveraged ETFs like TQQQ can have tremendously bad days, which you don't see looking at monthly numbers.
I enjoyed your explanation summary, too: "The shorter the sampling period, the fatter the tails."
The difference in the emotional experience between trading 3x Leverage daily and (roughly 3x Real estate using 25% equity) RE LPs is starkly different.
Yup. I was telling a friend this the other day - if you're going to put money in the stock market, you have to plan on holding for at least 10 years because if you randomly draw from a 10 year return distribution, you are very likely to harvest a great return; whereas if you randomly draw from a 1 year return distribution it isn't clear at all that things will go well.
Also, global diversification to avoid taking a country-centric bet. So $VT>$VOO. And some degree of asset class diversification to avoid an equity-centric bet. So VT + TLT (efficient frontier max sharpe) > VT alone.
And if you calculate kelly bet on historical equity returns, you'll realize 100% equities is already > 1/2 kelly. So If you reduce the volatility by diversifying into bonds you don't need to lever up to get the sharpe, you can just sit tight as the diversification moved you to a more tolerable kelly fraction.
it's included although you want to think in terms of CAGR which is 10.2% in the data i used (11.4 in a simple avg which isn't meaningful since it assumes you don't re-invest)
That makes sense. Just wondering if maybe we should then replace in the formula the 11.4% by the average after taking out all historical returns below -45%, and call rho the probability of a disaster above 45% to avoid double counting ? or call rho the "extra" probability of a crash ?
"Risk depends on the resolution". Great Comment. I invest in real estate LPs. They run 3-6 years and have no intermediate liquidity, and thus no perceived volatility. I review the financials, and operating reports, but there is no mark to market.
On the other extreme 3x leveraged ETFs like TQQQ can have tremendously bad days, which you don't see looking at monthly numbers.
I enjoyed your explanation summary, too: "The shorter the sampling period, the fatter the tails."
The difference in the emotional experience between trading 3x Leverage daily and (roughly 3x Real estate using 25% equity) RE LPs is starkly different.
Yup. I was telling a friend this the other day - if you're going to put money in the stock market, you have to plan on holding for at least 10 years because if you randomly draw from a 10 year return distribution, you are very likely to harvest a great return; whereas if you randomly draw from a 1 year return distribution it isn't clear at all that things will go well.
Also, global diversification to avoid taking a country-centric bet. So $VT>$VOO. And some degree of asset class diversification to avoid an equity-centric bet. So VT + TLT (efficient frontier max sharpe) > VT alone.
And if you calculate kelly bet on historical equity returns, you'll realize 100% equities is already > 1/2 kelly. So If you reduce the volatility by diversifying into bonds you don't need to lever up to get the sharpe, you can just sit tight as the diversification moved you to a more tolerable kelly fraction.
it's included although you want to think in terms of CAGR which is 10.2% in the data i used (11.4 in a simple avg which isn't meaningful since it assumes you don't re-invest)
That makes sense. Just wondering if maybe we should then replace in the formula the 11.4% by the average after taking out all historical returns below -45%, and call rho the probability of a disaster above 45% to avoid double counting ? or call rho the "extra" probability of a crash ?
but isn't the rare large drawdown already factored in the 11.4% average typical historical return ?