The Rules, Part XIX

There is room for a new risk model based on the idea that risk is unique among individuals, and inversely related to the price paid for an asset.  If a risk control model has an asset becoming more risky when prices fall, it is wrong.

After doing my talk for the Society of Actuaries last Wednesday, I got inspired to write something about modern portfolio theory, the capital asset pricing model, the efficient markets hypothesis, etc.  This particular rule deals with two things:

  • The same event can have different risk for different individuals.  Risk is unique to each individual.  It cannot be summarized by a single statistic for comparative purposes across individuals.
  • In general, with a few exceptions, risk is inverse to price.  As the price gets higher, so does risk.  As the price gets lower, so does risk.  The major exception to this rule is when trends are underdiscounted, because estimates of intrinsic value are flawed.

Let’s deal with these issues one at a time.  Start with a simple question.  Why do academics want to have a single measure for risk?  It allows them to write papers, and it keeps the math simple.  That’s why we have concepts like beta and standard deviation of total return.  It’s why we have concepts like the Sharpe ratio and other ratios that purport to measure return versus risk.

If our total planning horizon was similar to the periods that these figures are calculated over, they might have some validity.  But most of the time are planning horizons are longer than the periods that these figures are calculated over.  Even worse, most of these statistics are not stable.  The value calculated today may likely have a statistically significant difference from the value calculated a year ago.

But what is worse still is the idea that by taking more risk you will get more return.  If anything, the empirical research that I’ve been reading, and the value investors that I have talked to, indicate that the less risk you take, the more you’ll make.  A good example of that would be Eric Falkenstein and his book Finding Alpha.  Minimum beta and minimum standard deviation portfolios tend to outperform the market.  Junk grade bonds tend to underperform investment-grade bonds.

If it hurts too much, don’t do what I’m about to say.  Think about Lenny Dykstra.  When he and I were writing at RealMoney.com at the same time, I would often ask him about what his method would be to control risk.  He never gave me a good answer; actually he never ever gave me an answer at all.

My concern was for small investors, dazzled by the celebrity, and the simple approach that he would take that seemingly yielded huge profits, would adopt the approach, and not know what to do when things went wrong.  For Dykstra, who seemingly had a lot of money, losing a little on a deep in the money call trade would not hurt him much.  But to an unfortunate average guy reading Dykstra’s work, a similar sized loss could be very painful.

That said, that greatest risk was in plain view, which Steve Smith, I, and a few others went after — Larry didn’t know what he was talking about.

Risk varies by differences in wealth; risk varies with age.  Risk varies with the level of fixed commitments you have in life.  To give you an example there, when I went to work for a hedge fund, the first thing I did was pay off my mortgage so that I would feel free to take big risks for the hedge fund.  It is far harder to take risk, the higher the level of fixed obligations that one must pay month after month.

To make it more practical, think of all the malarkey that has been spilled talking about “animal spirits.”  I don’t believe that businessmen are irrational; many Keynesian economists are irrational, but no, not businessmen.  Businessmen will not take risks when they are overleveraged, or, when a broad base of their customers is overleveraged.

Risk is unique to everyone’s individual situation.  Any time you hear someone bring up risk factors that are generic, you can either ignore them, or, more charitably think that they have a proxy that might have something to do with risk, maybe.

Go back to Buffett’s dictum: far better to have a bumpy 15% return than a smooth 12% return.

-==–==-=-=-=-=-=–==-=-=-=-=-=-=-=-

The second part of the rule says that risk models should reflect higher risk as prices rise and lower risks as prices fall.  The implicit idea behind this is that it is possible to calculate the intrinsic value of an asset.  Can I disagree with one of my own rules?  Well, since I do the writing here, I guess I get to make up the rules about the rules.

There are many assets that it is difficult calculate the intrinsic value thereof.  Examples would include commodities, growth stocks, and anything that is highly volatile.

Though I believe my rule is correct most the time, markets are subject to momentum effects.  Often when a stock is at its 52-week high, that’s a good time to buy, because people are slow to react to changes in information.  And, when stock is falling hard, and is at a 52-week low, that is often a good time to not buy the stock, because there are maybe bits of information about the stock, or its holders, that you don’t know.

In general, though, higher prices are more likely to be overpayment and lower prices are more likely to indicate bargains.  Why?  Because returns on equity tend to mean revert.  Companies with poor returns on equity tend to find ways to improve business.  Companies with high returns on equity tend to find increased competition.

Thus, as always, I counsel caution.  Don’t ignore momentum, but also don’t ignore valuation.  Ask yourself how much upside there could reasonably be, and how much downside.  Play where the downside is limited relative to the upside, because the key to investing is margin of safety.  Play to win, yes, but even more, play to survive, so that you can play longer.