Gyroscopic Investing

I'm thinking about an allocation that might be appropriate for a highly risk averse 80+ year old.  The main goal is capital preservation as opposed to capital appreciation.  For example, the prospect of a 10-15% decline is really not very acceptable. 

Clive has suggested volatility can be reduced by mixing a PP 50/50 with a long/short treasury barbell, resulting in a net 37.5% allocation to LT and ST and 12.5% gold and stocks.  This does reduce volatility at the expense of some return - from 1972 through 2008 the PP average return is 10.1% with a stddev of 8.42% (and compound return of 9.8%) while adding the long/short barbell changes this to 9.37% with a stddev of 6.27% (and compound return of 9.2%).  I've looked at a 50/50 mix of PP and cash (short treasuries) - this reduces the stddev even more to 5.27% with an even lower yield, i.e. 8.9% (compound return of 8.78%).

My question is what is the general consensus about reducing volatility with either of these approaches?

With both of these the gold allocation is cut in half, which seems to imply although they're less volatile either of these might actually be MORE risky than the standard 4x25 PP since these portfolios won't respond the same way to (say) high inflation. Reducing volatility sounds nice, but on the other hand the potential for getting mostly wiped out due to high inflation or an Iceland-style currency collapse is really not very acceptable either.  I've looked at reducing the stock exposure, even all the way to 0 but this actually increases volatility and decreases return (kind of perverse, but makes sense).

Is the bottom line that the volatility of the standard 4x25 is basically the price you have to pay for the protective aspects of the portfolio?  It seems like there should be a way, given a willingness to forgo some return, to reduce volatility but keep the protection.  Thoughts?

I have been doing a lot backtesting of portfolios trying to tweak the PP. Something we have to keep in mind is that we are measuring volatility with nominal numbers instead of inflation adjusted. This makes gains made during a high inflation period look bigger than they are when using real numbers, and gains made during a deflation look smaller than they truly are. Therefore you have to make sure to not  rely on backtests using nominal numbers. You are right to assume that your tweaked PP is more susceptible to inflation, although it appears less risky when using nominal standard deviation as a measurement. I wish there was a backtesting tool that would allow us to use inflation adjusted numbers. That would be quite useful. Im guessing the PP would look even better when comparing it to traditional strategies volatility.

If you want to reduce volatility in just about any portfolio you are better off holding more cash. This will of course reduce the growth prospects of the portfolio. However, a high quality Treasury fund for your cash will probably be able to tread water during high inflation and will provide a little growth in other scenarios.

For instance if you had 70% in cash and 30% split amongst stocks, bonds and gold the portfolio averaged 7.1% since 1972 with no losing years. This is compared to the standard portfolio that is 25% split that had 9.1% growth and lost somewhere around 5% in 1981. The modified portfolio even provided modest real returns during the bad inflation of the 1970s (10K grew to almost 12K from 1972-1982 vs. 16K for the unmodified version).

We can't predict the future volatility of course, but holding more cash is generally the most reliable way to reduce volatility at the expense of returns (barring some type of severe hyper-inflation scenario).

One portfolio I happened upon was:
70% Total Bond fund
10% TSM
10% Gold
10% REIT's

It had a pretty competitive return (on SIMBA's spreadsheet it was 9.03% with the worst loss being 3% in 2009).  It beats the 80ST/10GOLD/10TSM by just under 1%...

It's something to consider...  slightly riskier, but still pretty darn safe.