Yesterday, I described a deadly psychological bias that has accounted for many of the trading woes we’ve seen over the past two years. In this companion piece, I’d like to outline a different cognitive bias that afflicts the majority of discretionary traders I encounter, with similar deadly results.
Imagine that August is a very hot month. We are preparing a trip for late October and I say to you, “Look at how much hotter August has been than June or July. Temperature has been trending for the past few months, so we need to be prepared for a scorching hot October!” No doubt, if you took my advice, you’d be stuck with me in chilly fall drizzle.
Of course, no one would seriously extrapolate temperature from August to October because we recognize seasonal patterns to temperature. In other words, we think in terms of cycles, not linear trends. There are portions of the year in which temperature trends higher; there are portions of the year in which temperature trends lower; and there are portions of the year in which temperature can oscillate in a range.
Thinking in cycles occurs in many contexts. For instance, in planning our days, we typically schedule periods of greater activity and lesser activity. We don’t simply assume that, because the morning was productive, we should tackle even more over the course of the afternoon and still more overnight. In our personal money management, we don’t assume that a good income period will expand forever and predicate our spending on income streams that we linearly extrapolate. Rather, we budget more modestly, understanding that the good times can be interspersed with lean ones.
The deadly cognitive bias affecting traders is the tendency to think solely in terms of trends. We look at charts, we draw trendlines, and we extrapolate those into the future. We seek directional movement, so we project direction.
Too often, the virtues of keeping trading simple lapse into simplicity. Just as trends seem poised to run, we add to our positions, and then experience reversals just when we’re most vulnerable. A few years ago, I conducted a study of portfolio managers in which we placed their trading returns into four buckets, representing levels of risk- taking. At the three lowest levels of risk-taking, the managers were consistently profitable. When taking the greatest risk, however, the managers actually lost money!
There are many possible explanations for the finding, including overconfidence bias. In the case of these managers, however, the overconfidence bias was itself rooted in a larger cognitive bias. The high risk taking occurred after a period of strong profitability. The managers had “caught the move” and, in a desire to make the most of the trend, added to positions. They kept adding until August became October and their returns became cold and wet.
Yes, such extrapolation could also be considered an example of recency bias: predicating future returns on the latest ones. More fundamentally, however, there is an epistemological bias at work: the use of trend as one’s unit of analysis.
Consider the following data exercise (raw data from Index Indicators). Over the past ten years, we will take each of the 500 stocks in the Standard and Poor’s 500 Index and calculate each day whether it closes above or below its 20-day moving average. We then aggregate the data to calculate the overall percentage of stocks closing above their 20-day averages. How does the percentage of stocks trading above their 20-day averages affect performance over the next 20 days? Please answer this question in your own head before reading below.
As it happens, over a ten-year period covering many bull, bear, and flat market periods, returns have been greatest in the top quartile of readings, when the greatest majority of stocks have been above their 20-day moving average (next 20-day average gain of +.70%) and in the bottom quartile of readings, when the greatest majority of stocks have closed below their 20-day averages (next 20-day average gain of 1.14%). When the majority of stocks were neither strong nor weak, returns were tepid overall (next 20-day average gain of +.01%): negative when the majority of shares were below their 100-day averages, positive when the majority were above).
What this means is that we have evidence of trend (broad strength leading to further strength) and evidence of anti-trend (broad weakness leading to strength). We also have evidence of cycles nested within cycles, as the outcomes were partly mediated by how stocks traded relative to their 100-day averages. In asset management parlance, we have evidence of momentum effects in markets and value effects–and we have evidence of those occurring across multiple time frames.
Think of it this way: any market can be modeled as the sum of a linear component (trend) and one or more cyclical components. When the linear component approaches zero, we have an oscillating, range market. In those situations, fading both strength and weakness become profitable strategies. When the linear component approaches +1 or -1, we have trending markets and are more likely to see strength and weakness beget more of the same. Many markets have significant linear and cyclical components, which means that it makes sense to buy dips in uptrends and sell bounces in downtrends.
All of that should make clear that any single trading strategy cannot work in trading or investing. The simplicity of trend trading and countertrend trading is too likely to devolve into oversimplification. Sticking with a single strategy when linear and cyclical components of markets shift dynamically–and then enshrining that inflexibility as discipline!–is a recipe for disaster.
Many, many, many of the emotional problems traders encounter are the result of a flawed epistemology. They are viewing the earth as flat when in fact it’s round–and so they never discover new continents. If more investors engaged in cycle analysis, perhaps fewer would need psychoanalysis.
This article was written by Brett Steenbarger from Forbes. This reprint is supplied by BNY Mellon under license from NewsCred, Inc.
BNY Mellon is the corporate brand of The Bank of New York Mellon Corporation and may be used as a generic term to reference the corporation as a whole and/or its various subsidiaries generally. This material does not constitute a recommendation by BNY Mellon of any kind. The information herein is not intended to provide tax, legal, investment, accounting, financial or other professional advice on any matter, and should not be used or relied upon as such. The views expressed within this material are those of the contributors and not necessarily those of BNY Mellon. BNY Mellon has not independently verified the information contained in this material and makes no representation as to the accuracy, completeness, timeliness, merchantability or fitness for a specific purpose of the information provided in this material. BNY Mellon assumes no direct or consequential liability for any errors in or reliance upon this material.
Brett Steenbarger, Contributor