Saturday, May 31, 2008

A portfolio for Extremistan

So friends ask me, what do you do for your portfoilio? Are there rules of the road in Extemistan?

It's more of an art than a science, but I have some intuitions that I follow and that have served me well. One is good return, and another is don't get cocky. Don't worry about betas, volatilities, and correlations. That stuff is highly questionable at best and outright BS at worst. Everything known for sure about price behavior strongly hints that their statistical distributions are "fat-tailed," subject to "large fluctuations." Moments like the variance are probably not even defined, being either infinite or at least very and fuzzily large - so much so, they might as well be infinite.

Average return over some fixed time period is a stand-in for cumulative return, which is the right thing to look at to see how an asset has performed.* The problem is that, cumulative return isn't enough. You're not going to buy an asset ten years ago, or whenever the cumulative return period started. You're going to buy it now. So is it worth it now? All you can do is figure out if the asset is now overvalued. "Value" investing amounts to no more than paying a good price or less for something. To check valuation, people use all sorts of numbers, such as the price-to-earnings ratio. My favorite is the price-to-book ratio. Ratios below two are very favorable. From two to four is okay. Above four or five, you're getting into overpriced territory.**

The final principle is diversification may be hairy, but it's worth it. But you have think more diverse than many brokers have traditionally, beyond the old trinity of stocks, bonds, and money market. Think of real estate (yes, it's still a good investment, if you pick the right type), and commodities and raw materials (yes, they're volatile, but diversify within the class).

And that's it. I'm with Schwab and settled on three of their funds with the best relative rankings by these criteria: a small- to mid-cap value fund, real estate (with holdings more in commercial than residential real estate, and a lot of international coverage), and an international fund (valuation a little questionable, but needed for diversification's sake).

POSTSCRIPT: Back to another recent excursion in Extremistan for a minute, the water crisis. A friend pointed out that just the assumption of stationarity (underlying probability distribution being unchanging in time) is an assumption, just like the Gaussianity (bell-curve-ness) assumption. And that's true. If there were definitive evidence of non-stationarity, by all means let's drop it.

But stationarity is a simpler and more primitive assumption than Gaussianity. We have very good reasons, based on everything known about "open" systems with "flow-through" (rather than "closed" systems with fixed totals), to drop the Gaussianity assumption. It's a more specialized assumption than stationarity and thus more likely to be wrong: so says Occam.

Even if stationarity happens to be wrong in the end, there's no reason to assume that the time variation of the statistical distribution is due to human activity. There are all sorts of more likely possibilities. There's a pernicious assumption that "open" systems should be stationary, and, if they're not, it's humans' fault, dammit. In the case of climate, we already know of decadal to millennial timescale changes due to solar variability; on longer timescales, due to the Ice Ages, continental movements and changes in the Earth's orbit and orientation in space. No need to invoke human involvement unless there's some other "smoking gun" that can't be explained better some other way.
* Cumulative return = (present value/initial value) = (1 + inferred average annual return)years.

** There are more sophisticated approaches, a particular favorite of mine being the flow ratio. But such detailed analysis of an asset is worth it only if you're investing in a single stock or other asset at a time.

Labels: , , ,


Post a Comment

Links to this post:

Create a Link

<< Home