Managing Money by Measuring Risk

THE TRUE MEANING OF RISK

Suppose I told you that it is possible to objectively measure risk in the stock market. For those of you who are pros, I know what youre thinking: "B. S.-NO WAY!" Now, assume that you were long the stock market as of October 9.1989. and I told you that the odds were then better than seven to one that the market would fail. If you had known these odds and believed them, would you have changed your investment strategy at all? If you were long at that point. I think you would have.

The fact is that there is a way to measure risk in the stock market in quantitative terms; there is a way to determine the probability of the market going up x% versus going down y%. Its not a "system"; it is a consistent method of gauging the likelihood that the current market trend will continue or fail. It is an approach that allows a speculator or investor to change his or her basic focus from the determination of "value," which is subjective and constantly changing, to objective risk.

But what is risk? When I started my career on Wall Street in 1966.1 knell a lot more about playing poker than about the markets, but I also knew that there were many similarities. Both require skill and luck, but more skill than luck. Both require knowing how to manage money so that even if you lose a few hands, youll still be around to play in the next one. And both involve exposure to the chance of losing money, which is the meaning of risk.

In my late teens, instead of making minimum wage bagging groceries. I made a decent income playing poker. I was good at poker because I knew how to measure and manage risk in the game. Instead of focusing on the size of the pot-the value approach-I only stayed in when the odds were in my favor: my focus was on the risk involved if I stayed in the hand. Risk involves chance, and chance involves odds. Odds take two forms: either those set subjective ly by a professional

163 164

oddsmaker, or those that are measurable according to probabilities based on a statistical distribution of limited possibilities.

In poker, odds are measurable, concrete, objective. For example, assume you are sitting at the right side of the dealer in a game of five-card draw poker with five players. With a $10 ante, there is $50 in the pot after the first round. If the first player bets $10 and everyone has called but you, then there is $90 at stake in the game-your potential reward is 9 to 1 (even though $10 of the $90 is yours, you have to consider each betting round separately in risVreward terms). Assuming that you have four hearts and want to draw for a flush (the chances of a flush being a winning hand are a minimum of 1.0037 to 1), your chances are I in 5.2 that you will draw another heart-your risk is 5.2 to 1. With a risk of 5.2 to 1 and a reward of 9 to 1, the risVreward for the current round is 1.73 to 1 in your favor. With a risVreward of 1.73 and a probability f .9963 that a flush will win, the adjusted/risk reward is 1.73 X .9963, or better than 1.72 to 1 in your favor.

If you employ this kind of strategy consistently, you may lose individual hands. You may even have a bad run of luck now and then, but you will win much more money than you lose over the long haul. I dont consider this gambling. Gambling is taking a blind risk. Speculation is taking a risk when the odds are in your favor. That is the essential difference between gambling and speculation.

Naturally, because my focus on risk had worked so well for me in cards, when I came to Wall Street I sought a method of objectively defining the odds of being right or wrong when speculating in the stock market. But when I asked pros whom I respected how they determined risk, I got chuckles and comments like, "You cant measure risk in the markets. Its not like cards. Its not a mathematical business. The market is a random walk game," or "The efficient markets theory invalidates risVreward analysis."

Instead of talking about measuring risk, they spoke about distributing risk, or even more often focused on "values," saying things like "Find value, buy value and hold it, and youll do well overthe long term." This advice went against my grain: I didnt want risk exposure unless I could objective!), determine that the odds were in my favor..

The advice of market professionals is more sophisticated today than it was in the late sixties, but

it is not substantially or predominantly different. Most professionals today think in terms of distributing financial resources according to some relative measure of performance or value. For example. Alpha and Beta are typical tools used in stock portfolio management. Alpha is a measure of quality which compares the performance of an individual stock relative to the market. An Alpha value of 1 means that the stock has, on average, outperformed the market by 1 % per month, so if the market moves up 10% in six months, that stock should move up 16%. Beta is a measure of volatility. A stock with a Beta of 2 should be up 20% when the market is up 10%, or down 20% if the market is down 10%.

-Most money managers buy stocks according to some set Alpha and Beta combination, plus other relative factors such as price-earnings ratios, book value, and

yields; and they call this "risk evaluation." But think about it. What do these measurements really have to do with risk? What do they tell you about the current trend of the economy or the effects of a change in Federal Reserve policy? What do they tell you about the likelihood that the market as a whole might be subject to a sustained and/or dramatic decline? The answer is: "Very little!"

I am by no means implying that these measures are worthless, far from it. But to use them as primary tools in speculation or investment assumes that value is an objective, static concept. Value implies evaluation, which means that individual human minds determine it. What something is worth depends totally on what individuals in the marketplace decide it is worth. Value can change and often does ... rapidly. Consider the case of Penn Central.

In early 1970, Value Line Investment Service announced that the company was "worth" $110/share (as measured by the value of its underlying assets) and that it was "undervalued" at $7share. By this measure, the stock price should have soared. It went to Sshare! The analysts who calculated the value of the companys holdings failed to take account of the fact that those holdings would deflate in value during a recession. They assumed that the markets standard of valuation would remain unchanged.

More recently, in June 1990, some analysts were saying that Citibank was "undervalued" at $24 1/2. The last time I looked it was at $14. What they assumed was that Citiban was too big to fail, that there wouldnt be a recession, and that therefore Citibanks previously bad loans would tum into good loans. All of their conclusions rested on the assumption of this occurring, and did not take into account the risk of a market downtum.

In a market downtum, the Citibanks, the Tmmps, and any entity that is highly leveraged against assets that depend on dollar appreciation through continued inflation are subject to decline and possible failure.

Now consider the concept of buying and holding "value." The problem with this view is that there is only one tme measure of value-the market. For example, IBM has been a standard for such "value." But if you bought "Big Blue" in January 1983 or later and held it until November 1989, the n you would have been losing money during the third largest upward stock market movement in this century! Why ever "hold value" by being invested in a declining issue? Why experience all that pain when you could have made money instead?

The problem with the conventional approaches to market involvement is that none of them address one simple and fundamental question: "What are the odds that the current market trend will continue?" In other words, what is the risk of being long or short in the current market? Alpha, Beta, value, yield, PEs. book value-all of these measures have merit, but only as secondary considerations when you have a firm grip on the most likely direction the market trend will go. What is first necessary is a primary standard to gauge the risk of market involvement in general. So, how do you measure the risk of being long or short in the stock market?

A REVOLUTIONARY APPROACH TO RISK ASSESSMENT

Answering this question eluded me for several years. In order to measure something, you ha ve to identify a quantitative relationship established by comparison to a standard unit reference value. In the financial markets, this presents a perplexing problem: How do you establish a standard to measure risk and determine the probability of success without being arbitrary and subjective in the process?

Markets arent like a deck of cards with a limited number of permutations possible; they are composed of individuals engaged in the pursuit of their own unique set of desires or values that are by their nature subjective-unique to the situation and frame of mind of each person. So it would seem that to gauge market behavior accurately, to predict the likelihood of success of any investment, you would have to be practically omniscient-be able to poll every persons mind simultaneously and be certain how they would react to coming events. It is impossible to predict the future of price movements with absolute certainty. The best we can do is deal in probabilities, so the question becomes: By what standard do you measure probabilities in the financial markets?

Any time you speak in probabilities, you are speaking of the odds of something occurring based on a statistical distribution of possibilities. Insurance companies use statistical data, such as mortality tables, to set insurance premiums. For example, the current odds of a 24 -year-old white female dying in New York are 50,000 to 1. The average life insurance premium for a $100,000 policy for this group is $100 per year. According to the statistical odds, the probable outcome is that the insurance companies will gross $500,000 in premium payments for every $10,000 they pay out to the beneficiaries of 24-year-old white females who have died. That is odds of 50 to 1 in the insurance companys favor-not bad odds management. It is no wonder that most well-managed insurance companies are so profitable (the ones that didnt fill their balance sheets with junk bonds and real estate).

There is nothing certain about the life expectancy of any one individual, but tha t doesnt mean it is impossible to define a standard with which to gauge the odds of an individual person living to a certain age within a given standard of health-insurance companies make such measures their business. The same type of reasoning can be app lied to the stock market. In 1974,1 began a two-year, intensive study of stock market movements dating back to 1986, that I keep current to date. What I found is that market movements, like people, have statistically significant "life expectancy" profiles that can be used as the standard for the measurement of risk exposure. Let me explain.

After I missed the October lows in 1974,1 started asking myself some questions so I wouldnt make the same mistake again, questions like: "What exactly is a trend? How high or low does it usually go? How long does it usually last?" Drawing on Charles Dows identification of the three concurrent market trends-the short term, lasting from days to weeks; the intermediate term, lasting from weeks to months; and the long term, lasting from months to years-I went back in history and classified every price movement in the Dow Jones Industrials and Transportation

(Rails) averages, logging their extent (how much they moved in percentage terms from the previous highs or lows) and duration (how long they lasted in calendar days) in statistical distribution tables.

Using Robert Rheas classification methods (along with some of my own re finements), I identified primary movements (the long-term bull or bear markets), intermediate primary movements (the legs that make up primary movements in between secondary reactions), intermediate secondary corrections (important intermediate movements that move contrary to the primary trend), plus some other less important classifications (the hardest part of this process was in distinguishing between secondary reactions and minor movements against the trend). The result was a set of statistically significant bell curve distributions for the extent and duration of all market movements since 1896 such as shown in Figure 11.1.

In a bell curve distribution, statistical samples tend to bunch around the median or midpoint of the distribution. For example, the current median extent for bull market primary intermediate swings on the Dow is 20%. Of the 112 bull market primary swings since 1896, 57 of them, or 50.89%, have reached an extent of between 15% and 30%, while the minimum move was 4.3% and the maximum was 116.6%. Twenty-five percent of the moves went to highs above 30%, and 33.04% of the