Bermuda: LOM Weekly Perspectives

The markets are currently behaving much as expected, so we can take a little time off to examine some investment issues. Back in the seventeenth century the smart heads of the time (e.g. Newton and Leibniz) came up with the idea that the world was an ordered place with natural laws that we could discover, understand and use to control our environment. As a result, the theory of divine will and inevitable fate got a thorough bashing. But some folks just wouldn’t give up the concept of perfect order in all things and sought to explain anomalies, discrepancies and nasty events as part of a grand design. This was an exercise in rationalisation that provoked Voltaire into having a ton of fun debunking the nonsense. In truth, few things in nature, society or financial markets function with the precision of clockwork. That belongs to the field of classical physics and celestial mechanics. Outside those two areas we have to face up to the reality of chance events and even occasional chaos. Most processes are stochastic (i.e. outcomes occur within probability limits) rather than deterministic. So, to forecast we have to figure out the probabilities. What’s more, some processes can sometimes break down into chaotic behaviour before resuming orderly patterns. Whether we like it or not, this phenomenon has been observed in nature, as well as financial markets.

Rational investors don’t like to play the ostrich, and intrinsically accept the notion of probability and risk, as do gamblers. What’s the difference? Investors are supposed to make a careful estimate of the odds of success, based on the best evidence and analysis, while gamblers generally do not make such a calculation. However, there is often significant variation among investors with regard to the accurate assessment of probabilities and attitude towards risk. One of the major problems is that most people abhor complexity and seek to impose ordered patterns on reality where none exist. We can’t really blame such folk. They need a map to guide them, even if it is a bad one. We have all seen those atrociously inaccurate maps drawn up in earlier centuries, which were, nevertheless, better than no guide at all. Let us say that you actually have a good map that has guided you well in the past, but the terrain has changed and the map is now leading you astray. The golden rule here is to dump the map and get a new one and not try to fit reality to the redundant map. If a trading system starts to give poor results because of changing market conditions then it is obviously the system that should be fixed because you can’t change the investing environment. Simple, rational and obvious, but it is remarkable how often investors hang on to redundant ideas and unworkable models in the face of contrary evidence.

The random walk theory of stock market prices holds that the best forecast of a change in price in a forward period is the recent change in price, plus a random element. In other words, all the available information is already contained in prices and nobody has an edge in beating the market. However, most of the recent empirical evidence fails to support the theory. This means that there are pricing inefficiencies in the market that allow profitable trading. It also means that technical analysis, contrary to some fundamentalist contentions, can be a useful tool. Market participants have memories of past prices and this influences their buying and selling decisions. In consequence, past price and volume information can be used to forecast future price action. Hence, there is a case for using technical analysis, but it is still possible to use inferior models and techniques. A poor reading of charts can result in imposing an order and meaning on the market that simply does not exist. Like a Rorschach inkblot test used by psychologists. The inkblot is a meaningless splat on paper but a patient will, nonetheless, see images in it.

One way that people can improve their probability assessments is to distinguish skill from luck. Being at the right place at the right time is decidedly lucky for an investor but has nothing to do with his or her skill. Probability measures and risk-taking behaviour, which worked in that particular context, are misleading and can't be generalised. It is likely to produce miserable results in a different environment. Doesn’t sound too complicated does it? But the evidence shows that even fairly sophisticated investors often fall into this trap. In a way, we can say that people are often fooled by randomness, come up with the wrong odds and take irrational decisions. But, happily, these acts also bring about market distortions and inefficiencies that provide exploitable trading opportunities for others. The first order of business in arriving at better probability measures is to separate information from noise. It must be said that, with the growth of the Internet and financial news channels on television, the noise level has increased substantially over the past few years. Granted, free and useful sources of information have multiplied too. However, let us posit the untested but reasonable contention that the noise to information ratio has actually increased. This can increase volatility but also trading opportunities.

A major factor that distinguishes successful investors from the rest is that they apply greater skill and knowledge in arriving at better probability measures. We have to indulge in a bit of quantitative lingo here. Some of the stylised facts of financial data are that large outlying observations occur with high frequency, large returns tend to occur in clusters and periods of high volatility are often preceded by large negative returns. It means that the probability distributions, adequate for the task, have non-normal characteristics with regard to their shape (skewness and kurtosis), resulting in fat tails (that is, the probability of the occurrence of unusual events is higher than under the normality assumption). As a consequence, we need to use non-linear methods that take account of changing volatility and we have to incorporate regime switching, as the environment changes. Basically, all that this fancy talk means is that if we have a wrong probability model of how financial variables behave we will make poor investment decisions – and a lot of people do just that. Now, let's summarise the discussion into a few rules: 1. Filter out the noise. 2. Build relevant models. 3. Get the probabilities right.