Gyroscopic Investing

... now with that blasphemic statement, I probably have your attention.  I felt like tinkering today.

I love the PP, and try not to tinker, but here's where I have a problem... the assets are so different in volatility.  I don't have a problem with the volatility itself, but moreso the variation between not just cash and the others, but even LTT's (St Dev =  12.41) vs gold (St Dev = 30.87).

To make things simple, I've been playing around with 40/30/20/10 (annual rebalance) PP's based on different priorities.

Through 2010, based on Craig's PP, the 4x25 allocation had a 9.84% CAGR and 8.22 St. Dev., and a -3.9% max drawdown in 1981.

Stability PP

For instance, if your main priority was Stability (and you were nervous about PP losses), a 40/30/20/10 (C,B,S,G) portfolio would have had a 9.03% CAGR, 6.09 SD, and -2.6% max drawdown in 1994.  2008 would have resulted in what looks like a max drawdwon mid-year of -6.6%, and the 70's faired just fine (despite such little gold and so much LTT's) with a 10.21% CAGR (1972-1980).  What would appear to be an inflation-prone portfolio returned over 10% during the 70's... why?  Since gold is so volatile, and LTT's aren't as much, it still did its job.  Does it really take 25% gold to protect yourself?

Growth PP

If you wanted to gear the PP towards growth based on historical returns, 40/30/20/10 (S/B/G/C), you'd have had 10.25% CAGR, 8.87 SD, and -5.58% max drawdown in 1981. 2008 would have been a near -3% loss.

Proper PP (Combination of growth and volatility considerations)

I don't particularly like those two approaches.  One seems way too safe (especially considering the outlook for ST interest rates), and one seems relatively risky for its benefits.  People often agree that the PP's weak spot is cash, especially today, but are too afraid of the volatility of the other assets to do anything but the 4x25.  The other assets seem to return better and are more fundamentally what you have driving your portfolio, but their inherant volatility can send the thing sharply out of wack without cash in some years (1981).

Well given these factors, if we reduce cash to the 10% portion of the portfolio, and then order the other 3 better assets in terms of stability of return in the 40/30/20 slots, could we actually improve the PP returns AND reduce volatility?  Let's see.

Since the SD of LT Bonds, Stocks, and Gold is 12.41, 18.58, and 30.87, respectively, the "Proper PP" will be 40/30/20/10 (B/S/G/C).  This PP results in a 10.01% CAGR (better than PP), St Dev of 8.19 (better than PP), and a Max Drawdown of (drumroll) -5.02% in 1981 (DAMMIT!).  Keep in mind, though, this gold/cash-light portfolio had a 13.3% CAGR from 1972-1980 (plenty, and had a 3.95% return during the 2008 crash,with what appears to be a similar max-mid-year drawdown to the traditional PP).

The max drawdown is a bit more in 1981, but otherwise some pretty good performance, especially considering the pathetic ST rates we're likely to see for years.

Maybe the other two (Stability and Growth) appeal more to those with certain feelings about how they'd like to invest and their stomach for volatility.

I just like to think of this stuff... in this interest-rate environment I see cash significantly lagging the portfolio for years.

In the end, I think it comes down to the fact that these assets are different in their volatility, and to create a stable portfolio, having gold and LT bonds equally weighted with each other and with cash is a bit unreasonable.  You can actually build a slightly better PP.

Implied in my post is that LTT's have had better growth than gold.  It was a close call when I calculated it once, and I can't remember which won out.

Clive,

I really don't like the 5-year ladder.  Why?  Maybe year's 3-5 actually have a good, solid price move in a recessionary situation, but it won't be nearly to the degree of LTT's.  I actually prefer the volatility of 20-30 year treasuries.

If you look at any day-long or month-long shock to the stock market, almost all of them are met with huge gains by LTT's, gold, or both... usually the one that doesn't jump stays about flat or does ok.

I'm simply not interested in using cash-like instruments as a hedge against stock-losses... they don't zig enough when stocks zag.  I just don't see 5-year ladders offering the delveraging protection needed... especially since years 3-5 are in the zone where the fed can pull yields so low.  They simply don't have that hold on 30 year treasuries.

I'm not just thinking of 2008, though that's a perfect example, and we could still be stuck at those levels.  I'm thinking of anything that will significantly threaten the long-term stock-market earnings is going to make LTT's, first and foremost, skyrocket.  I really don't trust 5-year bonds at 1.57% today to protect me against a 25%-35% drop in the stock market.  LTT's at 4.xx% have already proven they can do an excellent job of that.

To improve on things, I'd definitely consider some of the other choices you've mentioned in the past (value stocks, EM, small caps... I also like miners), but I don't think I'll ever tweak the PP towards bonds on the short end of the spectrum.  The 1970's have shown that rates can rise very quickly in a recessionary scenario, and the LTT's won't be too much of a drag on the portfolio (-4% worst loss).

The test of any two (or more) diversified investments' long-term CAGR with rebalancing is made up of 2 things. 1) CAGR of the individual assets, and 2) Correlation of the individual assets to each other.  I'll continue to bet that 1) long-bonds continue to yield more than short bonds, and 2) they will better zig when the stock market zags.

Nicely done, moda!  It's interesting to see where a thorough examination of relative volatility leads you in the PP.

Did you use Craig's data from here?  Was this along with the 2008 "fix" to use true 20-30 year Treasuries in 2008 rather than VUSTX the way that Simba's spreadsheet does?  I was wondering whether it'd make sense to "convert" the data from VUSTX to TLT from 2003 onwards to get better data on how a PP would have worked.

One concern I might have is that overweighting LT bonds would have been a little worse in the 70s than it would have been in the 2000s.  From what I understand, the 70s data is all VUSTX, which has shorter duration.  It might therefore make it appear that LT bonds are less volatile than they really are.  Would this change your ordering or weighting?  Or did you figure out some other way around this problem?

LW,

That is the chart alright, and I'm super glad you pointed out the difference... I didn't really know what he was using for the LTT portion all those years.

I suppose you could be right about LTT's in the 1970's.  The shorter duration might have been a problem.  I just don't see it being that much (15-30 vs 20-30?). 

This bothers me now.  One thing I just caught, too, is that today cash makes up 12% of the fund.  If that were also true in the 1970's that would have softened its losses even more.

LTT's are still the 2nd least volatile PP asset, and I'd imagine that if during the 1970's we'd have had a TLT fund, we'd still be seeing sufficient real returns.

LW,

Food for thought... due to the many years of heavy volatility and non-correlation between stocks and gold, a 50/50 rebalanced allocation between the two would have resulted in a 11.13% CAGR with a SD of 16.2.

Who'da thunk you could get 11%+ return from two assets that have returned about 9% and 10%, respectively.

The beauty of diversification.

moda0306 wrote: Food for thought... due to the many years of heavy volatility and non-correlation between stocks and gold, a 50/50 rebalanced allocation between the two would have resulted in a 11.13% CAGR with a SD of 16.2.

Who'da thunk you could get 11%+ return from two assets that have returned about 9% and 10%, respectively.

Absolutely wild.  I'd never be able to take a standard deviation like that for my own money but those are some impressive returns.