The limitations of Technical Analysis

Technical analysis studies price and volume trends and patterns to predict the market. It is based on the premise that market participants’ collective behavior is reflected in price movements. It believes that historical price behavior tends to repeat itself, and patterns can provide insights into future price movements. Common patterns include support and resistance levels, trend lines, and chart formations such as head and shoulders, triangles, or double tops/bottoms.

Technicians utilize a wide array of indicators and oscillators to identify overbought or oversold conditions, confirm trends, and generate trading signals. Volume analysis is an essential component of technical analysis that provides insights into the level of participation and enthusiasm of market participants.

The strength of technical analysis lies in the assumption of the collective wisdom of the crowd. Where technical analysis falls short is in its emphasis on a single variable price and its purely associational statistical approach.

The price setting mechanism in the markets is governed by the Auction Process. There are three dimensions to any auction: time, price, and volume. These three taken together provide greater context to how prices are being made and what types of market participants are making those prices. Under this framework, prices may be made under different conditions that can impact the lasting nature of those prices. By emphasizing price in most studies and indicators, technical analysis leaves out the context in which those prices were made.

Traditional statistical analysis is associational by nature. This means that statistical analysis aims to identify characteristics of a variable by sampling from a distribution of that variable. Simply put, if you have a jar of jelly beans and you want to learn about jelly beans sample a few jelly beans from the jar and you can infer some characteristics of all jelly beans from that sample such as flavor, smell, size, color, etc. In this example the market is the jar of jelly beans and if you sampled a red jelly bean that tasted like cherries you would reasonably infer that all red jelly beans are cherry flavored. This is associational statistical analysis. However, if there were other red jelly beans that were strawberry flavored your inference would not be accurate. By associating the taste with the color you may have found a strong correlation but you have not learned anything about the process that gives the jelly beans their flavor. To do so you would have to have a deeper understanding about how the jelly beans were made.

Causal analysis of the market recognizes that the observer must have some knowledge of the price-setting process which cannot be computed from the data alone or even the distributions that govern the data. Causal analysis takes associational statistical analysis one step further; “it’s aim is to infer not only beliefs or probabilities under static conditions, but also the dynamics of beliefs under changing conditions.”¹

It is well know today that correlation does not imply causation, but rarely is the next step taken. Technical analysis works just enough to keep you coming back to the table despite know the odds are not in your favor. Our brains are associational machines which makes it easy to fall back on static analysis. However, the world is in a constant state of change and we must be able to adapt to the change to survive. The associational mechanisms that kept our savannah-dwelling ancestors safe from predators are not applicable to the world we live in today.

¹Causal inference in statistics: An overview, Judea Pearl, September 2009, p.99

You can find this article and more of my writing at www.cedarshillgroup.com