The methodology doesn't seem to use the dividend effect (stocks rise before a dividend), or the January effect (calender year end dividends and January dividends are reinvested at a high point), or include yearly costs and fees and applicable taxes all which would bring the compounded interest down significantly. Also year averages can skewed in many other ways making calculation virtually impossible.
Most SP500 historical calculators seem to just use basic compound interest calculators adjusted for yearly yield average.
My point remains, if you invested 1000 in Gold in 1970, then sold in 1980 and bought stocks, then sold the stocks in 2000 for Gold again, and got out in 2010 and back into stocks you would be well off.
Gold leg 1 10yrs: 37.88 to ~600
SP500 leg 1 20yrs: 600 x 1800%
Gold leg 2 10yrs: 10800 x ~400%
SP500 leg 2 3yrs: 43200 x 58.8% = 68601 x (1000/37.88)
1.81 million. 1000 in 1970 was wages of an average summer job. 4 Trades and long term bullmarket timing. Silver en lieu of Gold wouldn't be much different either. This example is far from realistic, but its point is still valid. Even if you missed the timing somewhat but got most of the bull period you could have done very well. You could have even outperformed this example significantly.