3 schools of investment management: Fundamental, Technical and Quantitative
Many of our clients are familiar with fundamental, bottom-up investment management. In fundamental investing, an analyst evaluates a company’s stock on a number of quantitative factors (P/E ratio, earnings momentum, book value, agency ratings, current yields) and qualitative (company management, marketplace leadership, business niches, barriers to competitors, catalysts for change) factors. Analysis can focus on “bottom-up” analysis which looks at the company specifically or they can focus on “top-down” considerations which emphasizes macroeconomic themes. Fundamental analysis is a classic approach used by the majority of investment managers.
By comparison, Technical analysts spend no time looking at the company itself and focus all of their attention on the market data for the stock. Technical analysts mainly consider the market data of price and volume to identify patterns in market prices like the “head-and-shoulders” graph, or a price-resistance line. Technical analysis is also quite common and used, at least partially, by many investment managers.
A relatively new arrival is the Quantitative investment manager. These managers are driven solely by computer models and mathematical formulas. This school of thought identifies patterns, both technical and fundamental, that can be used to predict market winners and add value.
Quantitative analysis is used partially by many investment managers. However, for this white-paper, we would like to think broadly about Quantitative investment management as a school of thought. Thus, we will focus our attention here to those managers that only use quantitative analysis.
There are millions of useful patterns to consider when dealing with the economy or its sibling – investing - such as seasonality. Seasonality can reveal a long term trend distorted by a predictable seasonal affect. For example, economists know there is going to be a lot of hiring of temporary workers for the 10 year census report. Economists know that hiring picks up for retail stores during the Christmas & Holiday shopping seasons. They can exclude the effect of these short term hiring programs to determine overall long-term employment trends. For another example, families tend to take vacations during the summer when the weather is nice and the kids are out of school, so economists expect increased demand (thus, higher price projections) for gasoline in the summer.
These stories make sense. The causes and effects of these seasonal changes are measurable, logical and predictable. Our understanding of the world depends on understanding the pattern. For instance, no analyst would ever project a full year’s worth of retail sales just by looking at the 4th quarter of a year, during the holiday shopping season.
Investing is a science which is easily boiled down to quantifiable essentials, like the market prices of a stock. Therefore, it is one of the easiest sciences to derive patterns. Unfortunately, these patterns are easy to identify, but that doesn’t mean that they are more than random occurrences or simple coincidences.
Passing the test on arbitrary circumstances
“October: This is one of the peculiarly dangerous months to speculate in stocks in. The others are July, January, September, April, November, May, March, June, December, August, and February.”
– Mark Twain
We have previously highlighted (in our June 2012 newsletter, “Storm Clouds Again”) a 1987 study which determined a strong, predictive correlation between the S&P 500 and the price of butter in Bangladesh. Obviously, this correlation wasn’t really a pattern governed by logic; it was just a coincidence. The correlation was not causation.
Almost as soon as the pattern was discovered between the stock market and this price of butter, the relationship broke and investors had to go searching for the next magic bullet. They have plenty of magic bullets to choose from. For example, “Sell in May and Go Away” is a common British adage in financial circles. In 2012, that advice would have had you selling at the bottom of the market and missing the rally of the past 5 months. On the other hand, maybe we’re just fulfilling the prophecy of a different adage: the stock market goes up in most election years. Investment proverbs are neither rare, nor new.
What is new is quantitative data mining techniques have advanced to the point where programmers require no understanding of the logic inferred by the black-box output of their models. Quantitative black-box investing has made potentially illogical inferences even easier to act upon. The patterns divined by these models may not always be logical. The patterns don’t even have to be expressible. Computer programs crunch the numbers, follows instructions, but do not require any comprehension of their actions.
This is where due diligence comes in. There’s nothing wrong with quantitative investing. It simply requires analysts to check the logic of the underlying assumptions and the scope of the decisions.
Passing the test won’t always save you
For those readers who have kept up with our blog on westminster-consulting.com/Publications/Blog, you will know that I am a fan of the last year’s movie “Moneyball” starring Brad Pitt as the general manager of the Oakland Athletics baseball team. Here is a summary of the movie: the cash-strapped Oakland A’s had the same process for finding and paying baseball players as the richer teams. At the beginning of the movie, they don’t enough money to keep their star players. To compete, they change their tactics. Using quantitatively driven analysis, Oakland finds a new way to identify affordable players with high potential. The movie demonstrates the difference in quantitative evaluation of baseball players; the key metric for evaluating baseball players is their ability to get on base and, ultimately, deliver enough runs to win. Previously, Oakland relied on baseball scouts for their instinct, experience and intuition. Furthermore, the Oakland A’s – being a poor team – were forced into a “contrarian” strategy. They couldn’t simply buy the best players that their model identified. Instead, they had to buy cheap players that were disregarded (like a pitcher who throws the ball at a strange underhanded angle) or actively avoided (like a first baseman who suffered nerve damage).
Moneyball presented an insightful, plain-English examination of a quantitatively driven system, but I have to stress that the underlying logic of the system (i.e. winning baseball games starts by “buying” runs) is sound. This is key for evaluating quantitative systems. Does the model make sense? Furthermore, even if we assume the models do make sense, how much of a bet do you want to make on the logic? Keep in mind that good bets based on sound analysis can still go wrong.
Let’s look at a financially oriented example. Quantitative analysis suggests that municipal bonds have predictably higher prices and lower yields than treasuries. This historical pattern makes complete sense: income from municipal bonds is largely tax free, whereas income from US treasuries is subject to state and local taxes. Municipals can have a lower nominal yield than treasuries but have the same tax-equivalent yield to investors. Other than that, the ratio of prices between municipals and treasuries was essentially stable and often used as an indicator of relative value between the two asset classes. Quantitative managers, using their computer models, could trade between treasuries and municipal bonds looking for the best relative value. Quantitatively driven hedge funds were able to buy undervalued bonds (go “long”) with borrowed money raised “shorting” the overvalued bonds.
The supposedly indivisible relationship between municipals and treasuries has the benefit of being perfectly logical. It also has the distinction of being utterly false. The fear that widespread municipalities would fail during the 2008 financial crisis resulted in massively lower prices and increased yields. On the other hand, the overwhelming fear of a catastrophic worldwide market seizure resulted in enormous sums of money funneled towards US treasuries (i.e. a flight to safety trade), thus increasing the price of treasuries and lowering their yields. Very quickly, the relationship between municipals and treasuries broke down and computer trading models that depended on the stability of that relationship lost huge sums of money. Worse, some of those quantitatively driven hedge funds that borrowed against the overvalued US treasuries to buy undervalued municipal bonds collapsed.
Mistakes can happen anywhere and the best investment managers still had a rough 2008. Fundamental analysts were certainly not immune to the market damage, but at least they had the opportunity to exhibit foresight. In general, the 2008 financial crisis was bad for quantitative products because the stability of many established financial relationships broke down.
Computers simply do what they’re programmed to do. A computer model has no chance of knowing if sharp market changes represent a strong, but predictable, fluctuation or a sea change which necessitates a change to the model itself. Quantitative investing still has a place, but many quantitative investment managers who survived through 2009 have refined their models and may be a little more sensitive to major shifts in hitherto predictable patterns.
Holding out and hoping that the markets will return to a rational, previously historical equilibrium is not a guarantee of a winning strategy. In the words of John Maynard Keynes, “the markets can remain irrational longer a lot longer than you and I can remain solvent.”