Selling Strategies and Stop Losses
There are several ways to approach the issue of stop losses. Here are a few examples.
At one time we managed accounts for people. One of our strategies focused on Utilities. In this strategy, we created an index for each utility. The index combined current yield and projected dividend growth rate (determined by the historical pattern of the company). Once we computed the index for each company, we ranked the stocks by this index. Then we selected the ten that were the highest ranked for the portfolio. Once a quarter, we re-ranked the stocks and sold any that were no longer in the top ten and bought their replacements in the list. Over 15 years that strategy had a return of about 20% per year after commissions. It was counter-intuitive that a once-a-quarter adjustment could serve as an effective stop loss, but it did. However, we were dealing with utilities and the market was far less volatile than it is now..
I tend to think that strength ranking would be far more effective because it is based on what the stock is actually doing rather than on projections and dividend payments. By strength rank, I do not mean using the customary relative strength index (RSI) but that might work if it is based on 30 to 50 days rather than the typical 14 days. You could do what we did. We created our own algorithm that finds stocks that have a genuine “strength pattern” (the chart shows obvious persistent strength) rather than stocks that have had a bit of a surge during the last 14 days. Also, a once-a-week adjustment would probably work fine. Here’s how it would work. You sort the stocks in your database according to their strength measurement (not the RSI), with the strongest stock at the top of the list. The top 10 go in your portfolio. Each week or month you would re-sort your list and sell any stock in your portfolio that has dropped out of the top 10, and then buy the stock that has replaced it in the top 10. Here are some other ideas.
You could also use a “strength pattern” strategy but reinforce it with one or two others. For example, you could track the 5-day and 20-day simple moving averages of each stock. Donchian popularized this combination. Our tests show that simple averages are at least as effective as exponential averages. The 5-day average crossing below the 20-day average can be treated as your sell signal. This approach is sensitive and can stand on its own merit. You could plot those moving averages on your charts and if your stock either falls out of the top 10 (or whatever), or if its 5-day average crosses below its 20-day average, you would replace the stock with another in the top rank that has not had a sell signal.
Another way to give yourself a little more confidence in your stop loss strategy is to use the 4-week rule. Merrill Lynch researchers once tested a variety of sell strategies. The most effective they found was one in which they sold any stock that dropped below the lowest price of the previous 20 days.
For intermediate term investors, the author has found that a 4% rule works reasonably well. For example, at one time he conducted research for our traders at stockdisciplines.com in which he tracked the highest low price reached by a stock since its purchase. If the stock fell 4% or more below its highest low price, he sold it. The strategy worked reasonably well considering that it is a pure percentage-based stop loss strategy. However, this sell strategy is more suitable for people who have a high level of success at picking stocks about to rise. Even with only four positions in a portfolio, a 4% loss on one of them would impact the entire portfolio only 1%. William J. O’Neil, the founder of Investors Business Daily, likes 7% to 8% stops with a portfolio consisting of 8 or 7 stocks respectively. These combinations would also result in a 1% loss if the stop loss were triggered.
One problem with all these strategies is that wherever you place a stop loss based on a straight percent, you will probably experience more whipsaws than if your stop is based on volatility or relative ranking. We once tested the straight percent approach with stops ranging from about 4% to 20% and found a significant drop in whipsaws at about 15%. The problem with using 15% is that you would need a portfolio of 15 stocks in order to keep any loss (due to a triggered stop) at 1% of the total value of the portfolio. Ten positions would result in a 1.5% loss on each triggered stop. You have to decide for yourself what you can tolerate.
Finally, the best approach by far is to base your stop losses on the actual behavior of the stock and its own volatility pattern. Straight percentage stops are too rigid, and they are unrelated to the volatility of the stock. With volatility based stop losses, the stock itself makes the decision for you. It is also a sound approach statistically. When a stock moves below what is statistically probable based on its own pattern of behavior, it is sold.