Using moving averages to trade Starbucks

By Anthony Trongone, Ph.D.

This is a continuation of an earlier article (September, 2006) in response to assessing the performance of a 3, 5, 8, 13, 21, 50 and 200 day simple moving average. Although the investing literature is replete with articles discussing the efficacy of using moving averages as a predictor of future stock prices, across 643 trading days there was not a single success story – across all seven timeframes.

This article will begin by discussing if it is practical for us to apply moving averages as a way to enhance our forecasting ability. Before using moving averages, split your trading session into smaller segments by separating the overnight session from the regular session.

Starbucks, being in 37 countries, has a strong international presence. I travel to China frequently and have observed firsthand its growth and popularity in this region. Regardless of the country they are in, their pricing policy is relatively the same throughout the world. Since its split on October 25, 2006, this coffee retailer is up $6.135; however, this summer it had two disappointing earnings reports causing the stock to fall sharply from its $39.88 high on May 5, 2006.

This study investigates the previous 229 trading days, after SBUX 2-1 split on October 25, 2005 until September 20, 2006. In taking a 100 share long position at the beginning of each session, the overall profit in overnight trading session (OVS) was $9.045; whereas, the overall loss in the regular trading session (RTS) was $1.87. Such a wide performance discrepancy is noteworthy, especially since we are applying an 8-day simple moving average system during the regular trading session (9:30a.m. – 4:00 p.m. ET).


overnight trading session = (today’s opening price – yesterdays closing price).

regular trading session = (today’s closing price – today’s opening price).

moving average:

A moving average is the best known smoothing technique. Although the timeframe can be flexible, the more days in your analysis the less emphasis that is given to the more recent performance. Despite the timeframe, however, a simple moving average distributes the same weighting for each score.

Its predictive function is to follow emerging patterns of success or failure; however, since each score gets the same weighting; it is always below the actual price when they are rising, but above the actual price when they are falling. When rising prices begin falling or falling prices begin rising a moving average is slow to respond to these sudden changes; therefore, it is a “lagging” indicator.

monitoring performance

Before trading a system monitor its performance. In this case, SBUX is showing a profit of $7.175 in the 229 trading days (10/24/05 = $27.50 09/20/06 = $34.68). This is somewhat deceptive since the cumulative loss in the regular session is $1.87; therefore, we would anticipate suffering a trading loss when the system requires taking a long position; but, producing a profit when our system signals a short position.

How will an 8-day SMA distribute the $1.87 loss in the regular trading session?

After applying an 8-day simple moving average (SMA), the following table demonstrates the results of the two trading positions.

As you can see, in comparison to simply taking a 100 share long position, using an 8-day moving average exacerbates a poor trading performance. Either way, it does us a disservice, because on a buy signal or sell signal it is produces unprofitable results. Unfortunately, a long position in SBUX leaves us with $5.67 loss; whereas, the 98 short positions have shown a $3.80 loss.

The lesson is simple; don’t try to force a square peg in a round hole. Clearly, a moving average system does not work with this stock. Before applying a trading indicator run an analysis to accurately assess its forecasting ability.

tracking failure

The chart shows the actual running performance of taking a long position in SBUX using an 8-day SMA to make trading decisions. By monitoring its progress, we can review its performance. When it goes into a losing streak, it is a long slow process, with spurts of brilliance, such as the $4.30 run at the start of this year. Lately, it has been showing more promise, holding its cumulative loss around $8.00, before breaking upward. In fact, in the last seven days it has made $2.33.

In trading Starbucks using the short side of this system, its results are becoming more aggressive, as it has seen a series of very erratic price moves to both sides of the chart. In its five biggest unprofitable days (SBUX advancing in the regular trading session) the loss was $7.46.

Despite these results, there are specific ways that an 8-day simple moving average can be used to enhance the outcome when trading Starbucks. In my next article I will offer concrete ways to achieve this.