The Effectiveness of the 50/200 Dual Exponential Moving Average Crossover on the S&P 500

ABSTRACTThe subject of this paper is the evaluation of a simple mechanical trading system - a dual exponential moving average crossover on the S&P 500 index, with a shorter moving average of 50 days and a longer of 200 days (EMA 50/200). Although the system beats the S&P from 1990 to 2012, the buy-and-hold strategy has the upper hand in the longer time frame - from 1950 onwards. Still, given that most investors find it very difficult to maintain strict investment discipline for longer periods of time, the system and the S&P are tested on shorter sub-periods of ten years. Once the available data is divided into sub-periods, buy-and-hold emerged as the proper choice during a secular bull market, while the EMA 50/200 system was the superior option during a secular range-bound market. If we realistically assume that the average investor will find it nearly impossible to successfully time the beginning and ending of these markets, then buy-and-hold remains the best choice for investors with a very long-term investment horizon (longer than a few decades) - e.g. pension funds and young employees at the beginning of their careers. The EMA 50/200 system, on the other hand, is a more suitable option for investors who are past a certain age when they are not, financially and emotionally, willing to endure high volatility that comes with an all equity portfolio. The analyzed mechanical system allows them to enjoy the benefits of stock market exposure (capital appreciation and dividends) far longer than conventional asset allocation would suggest.INTRODUCTIONThe search for a trading system that can beat the stock market is probably as old as the market itself. Market participants have always been concerned with analyzing certain information in order to predict the future direction of the market. Whether it is the study of macroeconomic variables, financial statements of companies, graphic patterns of prices, insider tips or lunar cycles, the goal is always the same: find a system that will give a buy signal when stocks are "cheap" and expected to go up, and a sell signal when they are "expensive" and expected to start falling. Although the Efficient Market Hypothesis (EMH) and truckloads of research studies based on it later suggested that the search for a superior trading system is unnecessary and inefficient and that the optimal long-term decision that an investor can make is the purchases of the market itself (i.e. an index fund), the search for a system that "beats" the market is still on and probably will never stop. This quest is fueled by the fact that it is generally very difficult to empirically determine the (inefficiency of stock markets, and also by the logical deduction that numerous superior systems could exist out there of which the public is unaware because there are never published1.Although there are as many systems as there are market participants, they can be roughly divided into two types: fundamental and technical systems. Fundamental systems use variables that are in relation with the business activity of the company and its stock price. This is a broad spectrum including macroeconomic indicators, past growth rates of revenues, profits and dividends and the evaluation of their future direction, liquidity and leverage status, product and management quality and investment indicators such as e.g. market price/eamings per share (P/E) or market price/book value per share (P/B). A few fundamental systems are listed for illustration purposes. There is, for instance, B. Graham's concept of "net current asset value" from his classic Security Analysis (Graham and Dodd, 1934). Graham recommends buying companies which stock prices trade below their net current asset value (NCAV) per share - with NCAV being a rough estimate of the liquidation value of a company. Holding portfolios of these stocks should in the end provide investors with above-average returns relative to market indexes. Another example is the fundamental system popularized by J. …