Technical Analysis Works in Certain Situations
By and large, financial literature has virtually shunned technical analysis for being too simplistic, solely backward-looking, and easily exploitable. A reasonable argument is the rules-based nature of technical trading may translate to a decaying profit potential as more traders adopt similar approaches. However, this notion can be—and is—debunked by certain trades involving certain securities in certain timeframes. Specifically, the dynamic nature of price determinants and idiosyncratic drivers of traders’ behavior preclude one from judging precisely why a technical strategy works in some circumstances, but not in others.
The rationale of Using Charts
To be sure, technical analysis has secured its position within many traders’ toolboxes. Indeed, technicals allow a trader to quantify overall market sentiment. Additionally, it provides a gauge for the market’s consensus reaction to or anticipation of alleged fundamental catalysts— the key to picking entry and exit points. Although this is a valuable function of technical analysis, the primary rationale is to establish a set of descriptive statistics to judge price levels and movements—guided by past levels and movements.
Continuous Advancement of Technical Indicators
Moreover, this primary rationale is advanced by the application of sophisticated mathematics and statistics to price data. In recent years, a breed of model-driven trading strategies incorporating ever-increasing variables and automated trade execution emerged. The models emphasize the statistical nature of technical indicators. Notably, they aim to improve prediction accuracy by including adjustments to the mathematical expressions underpinning “vanilla” indicators.
Both the availability of intricately sliced price data (micro- and milli-second level, for instance) and the growing accessibility of algorithmic trading spawned a vast literature on profitable rules-based trading strategies and price/volatility models.
Quantopian and Quantiacs are dominant platforms in this field, having established communities populated by numerically savvy “quantitative” traders designing and testing statistical strategies for a chance to earn investment capital. These two websites provide learning materials including programming tutorials, backtesting procedures, and sample strategies. They are popular enough to host worldwide competitions and well-attended conferences. In fact, Steve Cohen—who frankly should not require an introduction—invested $250 million in Quantopian through Point72 Asset Management’s VC arm.
Quant Models Are Extensions of Technical Analysis
While I appear to digress, recall that the potent models I allude to are essentially extensions of the technical approach to trading. Price data is churned to unearth some benchmark—or produce a prediction—against which reality is measured, allowing one to assess whether the market is doing “what it should.” Modeling/sampling errors and personal biases unquestionably cap the reliability of such an approach, but the same applies to, say, fundamental analysis and the associated decisions of upcoming catalysts and their effects, appropriate future price based on a P/E ratio or DCF, and so on.
Quantitative Trading vs. Passive Investment Strategies
The relative attractiveness of the “quant” approach—which I stubbornly claim is an evolution of the technical approach—stems from the rise in passive investing (through ETFs) and newfound ubiquity of robot-advisors. Net inflows to passively managed funds stood at just above $100 billion in 2016, versus net outflows from actively managed funds of $150 billion over the same period. The direction of this trend has persisted since 2006.
On the other hand, the appealing fee structure of passively managed funds has led to them owning 13% of America’s stock market. Betterment, Schwab Intelligent Portfolios, Vanguard Portfolio Advisory Services, and Wealthfront—four leading providers of robot-advisors—have roughly doubled their AUM over the past year to $77 billion. 82% of new retail investments at these firms went into index funds and ETFs.
Common Denominator: Rules
Let’s forgo discussing the vocational implications to investment advisors. Note that at a superficial level at least, passive funds and algorithmically motivated “investors” operate through rules. The legion of technical strategies and prediction models available today do the same. Sure, the rules guiding these computerized market participants increase in sophistication and robustness with time. They even get better at thinking like humans, thanks to machine learning and AI.
Who said technical strategies can’t do—and haven’t done—the same?
While I am no expert in this area, I recommend checking out this example of an adjusted regression model that integrates the tendency of volatility to spike or recede “abnormally” in clustered periods.