Fortune Teller

Q&A with IFA: Market Timing with Moving Averages

Fortune Teller

Question: Some researchers argue that a market timing strategy based on buy/sell signals generated by a 50- or 200-day moving average offers a more appealing combination of risk and return than a buy-and-hold approach. What is your view?

Note: The question above was originally addressed in DFA’s Fama/French Forum. It set a record for receiving the all-time shortest response shown below:

“An ancient tale with no empirical support.”

Although it is not explicitly stated, we are quite certain that Professor Fama authored that incredibly terse response. Buying and selling based on moving averages is but one application of technical analysis, the attempt to forecast the direction of prices through the study of past market data, primarily price and volume. Unfortunately for its believers, all forms of technical analysis were rendered irrelevant by Eugene Fama’s efficient-market hypothesis (EMH) which states that current prices reflect trade information such as past price movements and volume levels. This, in fact, is the weakest form of EMH. A stronger form states that prices also incorporate all publicly available information, rendering fundamental analysis an exercise in futility.

Despite its rejection by the overwhelming majority of academic researchers, technical analysis refuses to make a graceful exit, and it’s easy to understand why. Aside from the fact that millions of dollars are made on the sale of software systems and instructional seminars, any market timing decision has a 50% chance of being right in the short term because the market goes up and down daily with only a slight bias towards the upside. Of course, this bias becomes more important over longer periods (measured in years or decades), but over shorter periods (measured in days or weeks), it is hardly noticeable. Thus, if we were to study a hundred different market-timing methods, we should not be the least bit surprised to find that one or two of them made six good calls in a row or made sixty good calls out of one-hundred. We expect this from chance alone. However, someone who is not so risk savvy could easily be fooled by randomness. Perhaps moving average and related momentum-based techniques receive the most attention because they have had the most success, but since there are so many timing methods up for grabs, we face a similar situation with active fund managers where we have only a few like Bill Gross that are household names, which shows us that luck is likely the primary explanatory factor of manager performance.

A recent academic paper1 out of Norway (but based on U.S. market data from 1926 to 2012) examined the real-life performance of market timing with moving average and time-series momentum rules (not to be confused with the momentum factor in a multifactor model of expected returns). Unlike recent past papers that touted the efficacy of these rules, the author of this paper (Valeriy Zakamulin) included the impact of transaction costs as well as the impact of the location of the split point between in-sample and out-of-sample data. One problem identified by the author is the popularity of the 10-month moving average rule over other periods such as 8-months or 12-months. Zakamulin notes that this is likely the result of data mining where all reasonable period lengths were tested and it so happened that the 10-month period gave the best results. This reminds us of when Weston Wellington of Dimensional Fund Advisors determined that stocks that began with the letter “M” had superior overall performance compared to the rest of the market. Zakamulin summarizes her team’s findings as follows:

“Our results cast doubts on the previously reported results on superior performance of market timing strategies. In particular, we find that the performance of market timing strategies is highly overstated to say the least…We believe that the myths about the superior performance of market timing appeared as a result of data-mining and ignoring of important market frictions.”

For some people, disposing of moving average and time-series momentum rules will be insufficient because they claim to rely on technical market indicators that are independent of current prices. They believe that these indicators provide an opportunity to exploit the behavioral deficiencies of other investors. One example would be the overall level of short interest as an indicator that the market is overly optimistic or pessimistic. Another example would be the volume of odd lot trades as an indicator of the prevalence of small investors in the market (usually taken as a bearish sign when they are buying and a bullish sign when they are selling). The academic paper2 that demolishes their hopes comes to us from the diametrically opposite side of the world, New Zealand. The authors examined the profitability of 93 technical market indicators. For each one, they went as far back as the available data allowed. In one instance, the price of a seat on the New York Stock Exchange, they went back all the way to 1820. As with moving averages and time-series momentum, the authors find no empirical support for the predictive ability of technical market indicators as measured by returns data for the S&P 500 Index.

“We give these technical market indicators the benefit of the doubt, but even then we find little evidence that they predict stock market returns. This conclusion continuously holds even if we allow predictability to be state dependent on business cycles or sentiment regimes…Overall, we do not find the market indicators generate profits that beat the buy and hold strategy.”

Their statement regarding business cycles and sentiment regimes addresses the possible objection that the interpretation of technical market indicators depends on where we are in the business cycle or the “market cycle”. The problem is that it is very difficult if not impossible to know where we are in a cycle at the present time. It only becomes clearer in hindsight.

To summarize, while we don’t realistically expect that market-timing based on technical analysis will ever go the way of cassette tapes, we do expect that it will become less and less popular over time as investors become increasingly educated. We at Index Fund Advisors will continue to do our part to make that happen.

1Zakamulin, Valeriy, The Real-Life Performance of Market Timing with Moving Average and Time-Series Momentum Rules (July 3, 2014). Available at SSRN:

 2Fang, Jiali and Qin, Yafeng and Jacobsen, Ben, Technical Market Indicators: An Overview (June 12, 2014). Available at SSRN: