Perhaps the most difficult decision traders have to make when creating a moving average is determining the length or period that best fits the situation. Regardless of whether you select an EMA or an SMA, shorter periods yield more signals, but a greater percentage of those signals are false, and longer moving average periods yield fewer signals, but a greater percentage of those signals are true.
One hitch: Signals occur later in longer-term moving averages than they do in shorter-term ones.
In general, the shorter your trading horizon, the shorter the moving average you want to select. For us, a nine-period moving average is nearly useless. It generates too many signals that are difficult to follow. More isn’t always better. You want your technical analysis tools to provide better signals, not more, because although getting good signals is important, avoiding bad signals is even more so.
Consider using a long-period SMA to monitor the health of a trend. Select either a 30-day or 50-day SMA, depending on the duration of the existing trend and prevailing economic conditions. If a trend has existed for a relatively long period of time, choose the 50-day SMA as an exit indicator. However, if the economy appears to be nearing a peak, tighten your exit procedures and shorten the SMA.
Traders can fall into a trap when trying to fine-tune the moving average — or any indicator, for that matter — for a specific stock or situation. Logically, testing many different moving average periods using historical data to find the one that generates the most profitable trades and the fewest losing trades seems right, and charting software packages enable you to do just that to your heart’s content.
However, you’ll soon discover that what worked when using historical data often fails miserably when trading real money in real time. This problem is well known to statisticians and economists who build mathematic models to forecast future events. It is called curve fitting because you are molding your model to fit the historical data.
Know that fine-tuning an indicator for a specific stock or index rarely has any predictive value, and you must avoid it. You simply can’t trade using historical data. You’re better off settling on a moving average period that satisfies the requirements for a great many situations, rather than trying to fine-tune the time frame of a moving average to fit each stock.