Tag Archives: stocks

In my weekend research report I write for my clients, I shared a few thoughts on correlation. This is a timely topic-we may be entering a period of the market history where a lot of assets link together. This is a risk, as there’s always the risk of getting smacked in the face when (previously) non-correlated markets snap into lockstep, but there’s another side to this, as well. Correlated risk can drive exceptional performance, for those traders who understand the risks, can fully accept those risks, and manage them appropriately. Here’s the snippet I wrote over the weekend: (click to enlarge) Correlation is a concept that is often misunderstood and much-abused in thinking and writing about financial markets. This week, we need to devote a few words to the concept of correlated risk and how to manage these risks, but, first, a few thoughts on correlation and some potential issues with the measure: Think carefully Correlation is well-understood mathematical measure, but, as in so many cases, applications to financial markets can often encourage sloppy thinking. On one hand, perhaps too much attention is focused on correlation, and this is likely a side effect of correlation’s importance in constructing portfolios; expected future correlation of assets is one of the key inputs in portfolio models, and here we arrive at one of the first issues. Few people, outside of those who have done quantitative work and study on financial market data, appreciate how variable correlations can be. Stocks are put into portfolios with the idea that they are “high beta” or “low beta” (beta is a measure of the amount of a stock’s movement that is attributable to the broad market), but, depending on what time window we measure, a stock’s beta might fluctuation between 2.0 and -1.0 within the course of a few years. Even supposedly well-established rules of thumb, such as the Dollar’s correlation to other asset prices, stock markets’ volatility’s (inverse) correlation to price movements, are highly variable. Furthermore, correlation does not show what most casual market participants might expect. It is easy to look at two lines on the chart that go in roughly the same direction and say “they must be highly correlated.” Maybe, but maybe not. Correlation can be highly deceptive when extrapolated to other time periods; it is trivial, for instance, to construct two price series that are perfectly -1.0 inversely correlated, yet both “go up” over a longer time period. (See this blog post for an example.) Last, two points that will be familiar to readers with a mathematical background, but that bear repeating: correlation does not imply causation. Every statistics student learns this in the first class she takes, but we are all vulnerable to drawing unsupportable conclusions in the heat of battle; we cannot hear this reminder too often. Second, correlation measures linear relationships. In many cases, linear correlation has application to non-linear relationships, but just be aware that relationships between markets could potentially be much richer and more complex than could be captured in a measure of correlation. Correlated portfolio risk These points are interesting (and important), but the practical issue of correlated risk demands attention. Put very simply, we put on different positions in a trading book or portfolio with the idea that they are all independent bets-some may win, some may lose, but they will each “do their own thing.” Correlated risk is the risk that, for some reason, all of these positions move in the same direction at the same time. Why might this happen? It is well-established, through decades of market history, that correlations tend to move toward 1.0 in times of market stress . These are not rare events; they happen several times each decade when supposedly unconnected assets all move in the same direction, usually led by mini-crashes in stock prices. In retrospect, there’s always some “obvious” connection, but psychology is the driver. Market participants become scared and basically scramble for the exits all at the same time. However, risk is a double-edged sword, and one of those edges works for us. Risk is the companion of opportunity, and, in fact, we can reduce our job to taking on the right kinds of risk and managing them appropriately. Over a long period of time, many traders find that their returns tend to be concentrated in certain time periods; this is true both of discretionary traders and of trading systems in test and design. During these periods, risks tend to high and correlated, but there is an important lesson here. Accepting these correlated risks can lead to outsized returns. We may wish it were otherwise, but trading and investing are uncertain games of chance. We may wish for steady returns and a rising equity curve that looks like a straight line (or an exponential curve), but reality is messier. Reality is periods of feast and famine, and correlated risks typically are high in the rich periods. We should be aware of these risks and manage them, but we should see them for what they are-a real risk with real potential rewards and, potentially, a major driver of potential outperformance.