Past performance does not dictate future returns. But the graphs in this link highlight the value of holding stock longterm.
http://allfinancialmatters.com/2012/08/29/sp-rolling-total-returns-1-5-10-20-25-and-30-years/
the S&P 500 has never suffered a loss in a 20-year period. Only a few 10 year periods have had a negative return.
Please see the chart above I posted titled 'The Stock Returns Never Average' above. It matters very little what rolling returns are (and these are nothing but averages over longer and longer time periods). Why? The markets are fat-tailed, and fat-tailed statistics has different properties from Gaussian statistics. One of the properties is that past averages have no bearing on the future (in a Gaussian world past averages are stable and future returns will converge to them eventually). So just because S&P only had a 50% loss doesn't mean it can't go down 60%. And just because the longest recession was 10 years doesn't mean it can't be 15. And 'never suffered a loss' is not particularly useful if you are losing 50% of your portfolio (and this would have a huge impact on your total and annualized return). Timing matters when investing in risky assets, so when you just accumulated say $5M and the market goes down 50% this will significantly impact the total return/annualized return, and even though over 20 years you might suffer no loss (by the way it could be 25 years, and 30 years as longer/deeper recessions come about, which is only a matter of time), the net result would be catastrophic for those holding 100% stock portfolios hoping they'll recover (and getting hit by another longer recession with a deeper bottom). Fat tailed markets simply have more degrees of freedom than Gaussian ones, and so things that Gaussian statistics says is impossible is very much possible (and not only that, but also bound to happen eventually). So to me the only thing that matters is limiting the downside, and getting whatever return that one can get with a barbell allocation. Historically this has worked well (except during the times when interest rates were low, due to artificial government involvement).
Vanguard's own retirement distribution calculator shows that a 50:50 allocation is more stable over longer periods of time (30-40 years) than 100% stock allocation (when no more contributions are made). Also as you get older, subsequent contributions have less and less impact on the overall portfolio, so Vanguard's analysis applies much earlier than distribution. Bottom line is that once we take into account real risk that the market statistics shows us, in order to manage long term risk to a large portfolio, the traditional age-based asset allocation is significantly flawed, and most people's understanding of how long term risk works is also flawed, which can be easily demonstrated with basic mathematics (even Gaussian math works for that). Risk increases with time, and no amount of averaging will be able to hide it, so in light of that, I believe that only a risk-based approach is the most prudent long term strategy vs. target date or age-based approach.
Leave a comment: