We’ve all seen the analyst on CNBC explain with a sense of authority that the S&P 500 index has just crossed above or below an important moving average (MA). We are led to believe this foretells a trend reversal but we never really see data supporting that idea after the statement is made.
This leads to the logical question, “are analysts talking about MAs simply to have something to say or are MAs useful?” The answer, as so many answers are in the financial world, is that “it depends.” Sometimes MAs work spectacularly well. This was the case in 2008 when proponents note they warned of an approaching bear market.
The chart above shows what many analysts remember. The 200-day MA provided a clear sell signal on December 31, 2007. Selling on that signal would have avoided the bear market which resulted in prices falling by more than 50%.
While it’s easy to see the value of that sell signal, looking closely at the chart, we can see the problem with MAs. There were a number of false signals prior to the one that worked so well. The final cross below the MA was the fifth sell signal in less than three months. Then, there was another losing trade in May 2008.
Losing trades are common for traders relying on MAs. Testing shows signals based on a long-term moving average, the 200-day or a 50-day MA for example, will identify losing trades about 60% of the time. The winning trades tend to be large and most losing trades are small. In the long run, the indicator generally seems to be beat the market. But, that outcome largely depends on the start date of the test period. From 1973 to 2000, a 200-day MA strategy would have beaten the market. But from 1982 to 2000, the MA underperformed the market. If you want to trade MAs you need to ask if you could stick with a strategy, in real time, that failed to beat the market for 18 years.
Another problem with MAs is that they will almost always give buy signals well after the uptrend has begun. This can be seen in 2009 when the 200-day MA gave a buy signal after the S&P 500 had gained more than 40%.
In real time, could you wait three months as the market moved almost straight up for a buy signal? This is difficult to do and this is one of the reasons MAs are difficult to follow. But, the delay in signals is partly due to the design of the MA.
MAs are a popular technical indicator because they are simple to calculate and can quickly highlight the direction of the trend.
To calculate an MA, you find the average of prices for a period of time. For example, the 200-day MA is found by summing the closing prices for the last 200 days and then dividing the sum by 200. The calculation moves with the market action. The next day, the oldest piece of data is discarded and the calculation is repeated with the 200 most recent closing prices.
If the most recent closing price is above the average, the trend is up. A down trend means the price closed below the moving average. Examples of this can be seen in the charts above. Because the calculation is using historic data, there is an unavoidable delay in the trading signals.
From the charts, it is easy to see that some of the trades would have delivered large gains or avoided large losses. As we saw, however, there are many other times when prices quickly move above and below the moving average. These are known as whipsaw trades, and are impossible to avoid with any MA. Testing generally shows most MA signals will result in whipsaw trades.
In addition to identifying the direction of the trend, MAs can also be used as a trading system. This can be done by buying when the price crosses above the MA and selling when the price breaks below the MA. This will keep traders on the right side of very strong trends, as it did in 2008. But, at times, it will underperform for long periods of time and result in a frustratingly large number of losing trades.
To overcome these problems, some traders use two MAs. For example, they might combine a 50-day MA with a 200-day MA. In this case, they buy when the short-term MA (the 50-day) crosses above the long-term MA (the 200-day MA). Testing shows this doesn’t solve the problem, it simply changes the day the signals are given. But, even with two MAs or with more complex calculations of the MA, there will still be whipsaw trades and lags in signals.
A less common use of MAs is as a filter for other trading signals. For example, it’s possible to combine overbought/oversold indicators with MAs. One popular strategy is to combine the 200-day MA with the 2-period RSI (relative strength index).
RSI is among the most popular technical indicators. It is commonly calculated with 14 days or 14 weeks of data. Using just 2 days makes the indicator more responsive to the market action. This is shown in the next chart where the 14-day RSI is shown in the center and the 2-day RSI at the bottom.
The 14-period RSI is fairly flat while the 2-day version fluctuates frequently between overbought and oversold extremes. One strategy is to buy when the 2-period RSI is oversold, with a reading below 20, if the closing price is above its 200-day MA. This strategy offers a clear set of rules for buying pullbacks in an uptrend. It can also be described as a quantified way to “buy the dips,” a popular strategy among individual investors.
The dip is defined by the 2-period RSI. When the indicator is oversold, prices are dipping. The MA defines the trend and maximizes the chance of buying a dip rather than trying to pick a bottom in a selloff. If prices are above the MA, the weight of the evidence points towards the pullback being a dip in an uptrend. If the price is below the MA, the market is in a downtrend and buying at that time is riskier.
MAs are useful as confirmation signals but many traders will probably continue using them as a timing tool despite the problems with that approach. MAs, and the problems with MAs, have a long history. One of the earliest references to the indicator is a chapter on “automated trend lines” in Technical Analysis of Stock Trends by Robert Edwards and John Magee. In 1948 they wrote:
And, it was back in 1941 that we delightedly made the discovery (though many others had made it before) that by averaging the data for a stated number of days…one could derive a sort of Automated Trendline which would definitely interpret the changes of trend…It seemed almost too good to be true. As a matter of fact, it was too good to be true.
Despite the acknowledgement that the idea was too good to be true, traders have hung on to the hope that MAs will work. The truth is they are great in hindsight but difficult to follow in real time. Difficulties arise from the high number of whipsaw trades and the lag time in signals. But, they can be effective as filters for other indicators. This might be their best use. To do this, take buy signals of other indicators only when the close is above the MA and ignore buy signals when the price is below the MA.
Using an MA to confirm a trade signal won’t get you on CNBC but it could increase your profits.