A New Anomaly: The Cross-Sectional Profitability of Technical Analysis
- Yufeng Han, Ke Yang, and Guofu Zhou
- A version of the paper can be found here.
In this paper, we document that an application of a moving average strategy of technical analysis to portfolios sorted by volatility generates investment timing portfolios that often outperform the buy-and-hold strategy. For high volatility portfolios, the abnormal returns, relative to the CAPM and the Fama-French three-factor models, are high, and higher than those from the well known momentum strategy. The abnormal returns remain high even after accounting for transaction costs. Although both the moving average and the momentum strategies are trend-following methods, their performances are surprisingly uncorrelated and behave differently over the business cycles, default and liquidity risk.
Return data are obtained from CRSP. The time period analyzed for the trading strategy is from July 1, 1963 through December 31, 2009.
Technical analysis techniques have been utilized widely by many traders and hedge funds over the years, but academia has been loath to embrace its benefits, choosing rather to attribute its effectiveness to blind luck, minor market nuances, or brief periods of irrationality displayed by investors. This debate has continued unabated since the sixties when proponents of the Chicago School of Business put forward their “Efficient Market Hypothesis”, or “EMH”, to dispel any notion that analytics can divine future pricing behavior. While most academic papers favor these precepts, the accompanying treatise entitled, “A New Anomaly: The Cross-Sectional Profitability of Technical Analysis”, supports the counter position in the ongoing debate.
Technical analysis has evolved over the past few centuries into a professional art form that is the foundation of trading systems far and wide, from the individual trader to the enormous hedge fund or international bank that attempts to benefit from market fluctuations in price for securities, commodities, and currencies.
These techniques rely on three principles:
- Price is a comprehensive reflection of all forces in the market;
- Pricing behavior over time is repetitive, and
- Price movements tend to follow trends.
Academia, however, has never accepted these principles, but has preferred to support their various theories on how a “market” operates and determines price. These theories include “EMH”, but also random walk theory, the Capital Asset Pricing Model, Modern Portfolio Theory, and the familiar Black-Scholes option-pricing model. The bursting of the Internet “bubble”, the near meltdown of the credit default swaps market, and the “Flash Crash” of 2010 are but three reasons to question the validity of these basic academic tenets. These models depend on markets acting rationally and that many variables are normally distributed, based on statistical approaches that have stood the test of time. Market “crashes”, however, stem from what happens outside of two standard deviations from the mean, the so-called “fat-tail anomaly”.
Regardless of criticism, many have chosen to believe in technical analysis techniques over the years. One need only review the success of quantitative hedge funds to know that gains can be gleaned from market volatility, inefficiencies, irrationalities, or whatever name one wishes to assign to the “phenomenon”. Traders understand that these techniques fail on many occasions, but with the right circumstances and at the right moment, “TA” can seemingly provide an “edge” for the trader, when that is all he needs to win consistently over time.
Eugene Fama and Fischer Black, major proponents of the Chicago School of thought, actually departed from the sanctity of their academic surroundings to design quantitative trading models for the likes of Goldman Sachs. EMH and option pricing were their respective “babies”, and each purported that price is based on all public knowledge, past and present, and that future price movements, regardless of statistical trend and momentum indicators, are nothing more than a coin toss, highly unpredictable as a “50/50” proposition.
Once on Wall Street, each man began to moderate his view to believing that the market is “partly” efficient. Black was heard to remark, “The market is more efficient on the banks of the Charles River than the banks of the Hudson,” duly noted once he was part of the Goldman Sachs team. Quantitative “black box” algorithms have continued to leverage brief periods of inefficiency and investor irrationality to make favorable returns, a potential rebuke of academic thought, but have yet to have “proof of concept” demonstrated in a broad-based, repeatable, back-testing test scenario.
Very few academic papers in the recent past have chosen to deviate from accepted dogma, but the accompanying paper seems to choose the contrary opinion and demonstrate a measure of success with technical analysis. Tests performed with a simple “10-Day SMA” trading system with NYSE/Amex stocks segregated into ten volatility classification portfolios delivered returns in excess of a “buy-and-hold” strategy over a period from 1963 through 2009. Additional tests were conducted based on size, length of SMA, liquidity, business cycles, and sub-periods, with comparisons to other momentum strategies, but the volatility portfolios approach was the most productive.
Yufeng Han, Ke Yang and Guofu Zhou investigated an SMA trading strategy over a broad base of stocks, segregated by volatility, and then compared results with the same stocks as if they were bought and held. They focused on a 10-Day Simple Moving Average, but also tested other longer period SMA’s. Their trading system followed these rules:
- When a stock’s price exceeded its 10-Day SMA, then it was purchased at the next day’s opening. If not, the funds were invested in risk-free T-Bills;
- When the stock’s price crossed below its 10-Day SMA, then it was sold at the next day’s open.
The results depicted in the above chart are “positive” returns over and above the Base “buy-and-hold” strategy. Adjusting for risk improves the performance, and, even after deducting estimated trading costs, the net returns are significant. Holding periods ranged from 10.5 days for the “Low” portfolio, to 7.4 for the “High”. Longer SMA’s, which resulted in longer holding periods, also produced substantial returns over time.
- Identify the top 10% of firms with the highest volatility based on the previous year’s daily return standard deviation–form a portfolio of these names.
- Calculate the 10-day simple moving average of the high volatility portfolio.
- Buy or continue to hold the portfolio when it breaks the 10-day moving average, otherwise, sit in cash.
- Make money.
There was little correlation of these results with momentum strategies, but more work in other areas is recommended. These results may also be questioned due to the timing of buys and sells. Pricing behavior can be affected by timing at closing and opening market situations. The authors also estimate transaction costs at a “fixed” 25 basis points, but high volatility stocks tend to suffer from wider spreads and lower liquidity, thereby costing more to transact than the study assumed. Some of the transaction cost issue would be lessened if the MA (10) strategy was applied in forex trading, where transaction costs are minimal, the performance of company management is not applicable, and diversification is easy to attain.
The findings of this study suggest what many technical traders accept on faith. There is opportunity in volatility, even with the simplest of trading modalities. Fluctuations in pricing behavior reflect a period where the participants in a market are attempting to consolidate on a new equilibrium price. While this “cogitation” process operates, speculators may find pricing inefficiencies and the opportunity for gain by using technical analysis techniques that provide an apparent advantage when only an “edge” is required.