The Limits of Technical Analysis, part 1

from CSS Analytics blog:

My early education was quite typical: I went to business school to get my MBA in Finance and then tackled the CFA designation. I was typically instilled with the “Efficient Markets Hypothesis” and often could be found reading arcane issues of “The Journal of Portfolio Management.” That was also a distant time when I thought CAPM was the single greatest equation in finance. I started my career many years ago in equity research, and then moved on to a stint in wealth management and investment advisory. I had the unique benefit of spending most of early years without looking at technical analysis. I spent years doing both macro-economic and company research, and also invested heavily in research using fundamental scoring systems. My early heroes were Joel Greenblatt, Michael Price, and David Dreman–all pioneers in original fundamental research. While it may be unbelievable to loyal blog readers, I distinctly remember the days when I used to scoff derisively at anyone who begged me to look at a stock chart! Now, after becoming a technical analysis convert so to speak, I have a fair degree of perspective to share on what technical analysis can and cannot accomplish.
First, let me state a unique premise: without other market players believing in fundamental analysis, I believe that technical analysis would not work. Huh? Well the truth is, all major trends including both bull and bear markets are a function of the belief in some theory about fundamentals. Few individuals with serious wealth are willing to risk their own money on the basis of an interpretation of a chart pattern. Imagine telling a client that has given you $25 million to invest that you are going to be investing $5 million in China Telecom because “the chart looks good.” Or how about trying to tell that same person that you are going to increase your exposure to Citigroup in 2008 “because my indicators show that the stock has hit a bottom.” You better have a good reason to explain to this same person why you are risking their hard-earned money on a bunch of chart squiggles. For this reason, most of the money in the mutual fund and pension fund universe is invested using fundamental research and macro-economic theses about a specific stock, sector or commodity. Typical examples include statements such as: “the world is running out of oil”, or that “the internet is going to take over brick and mortar business.”
Not some, but nearly ALL major bubbles or parabolic moves are created by feedback loops that start with a few people believing in a given theory and end with the majority achieving a consensus that a theory is true.
One implication of this process is that the proximity to a long-term top or to a bottom can be better approximated using gauges of sentiment rather than classic technical indicators. Based on the theory postulated above, major moves will start occuring when the majority of people do not believe that a market will reverse course based on fundamental reasons. The smart/more informed money will often be placing their bets ahead of the crowd without moving the market. By extension, this also means that major turning points will be difficult to distinguish since the money flow into a given stock or market will likely be offset by the majority of investors betting in the opposite direction. No technical indicator in my experience is capable of telling you with any accuracy where the top or bottom is. I have never tested or seen anything that can accomplish this significant feat of prediction. Strangely, it seems that most newsletters and market pundits spend most of their time calling tops and bottoms. Perhaps this is why they always seem to underperform buy and hold or a simple 200-day moving average rule despite having arguably excellent knowledge of a wide variety of truly predictive indicators.
The reality is that the fallability of technical indicators is a major limit of technical analysis that cannot be addressed adequately by trying to be more precise. The sad fact is that this never-ending quest for precision is the undoing of many great chartists. Many technical analysts fall prey to “confirmation bias” after analyzing a market because they are desperately looking for some “non-confirming divergence” or other magical signs that they are still correct in their analysis. This same problem also affects quants that trade mechanically: no hidden mathematical transform of price and/or volume (or combination thereof) holds the key to predicting all markets with significant accuracy. At some point no matter how sophisticated you are in your analysis, you will reach a maximum bound of profitability that can be achieved by looking at prices and volume in isolation. This bound can only be surpassed by incorporating variables that are not multi-collinear with price-based indicators. Not only is it easier to improve your precision with other variables, but you also run a much lower chance of having a poor reward to risk ratio.
to be continued………
original story