My research shows that diversification (holding a
larger number of Select funds) using the Select Switching System
tends to have two main effects: 1) return is reduced, and 2) volatility
is reduced. When two or three funds are held, the reduction in
both is moderate. As more than three funds are held, the reductions
become substantially greater, primarily because of time spent
holding the money market funds. There is no right or best number
of funds since each investor has his or her own investment horizon
and risk tolerance. Personally, I have let all my tracks combine
into a single Select fund. However, I have a fairly high tolerance
for risk and long investment horizon (at least ten years).
Increased volatility after 1996 has caused me to re-think the above. The following is an excerpt from my commentary dated June 15, 1998:
It is apparent that volatility of the market as a whole and sector funds in particular has been increasing since the end of 1996, and this is shown by statistical measures such as the standard deviation of the average weekly changes of the Select funds. Since markets and individual stocks and mutual funds tend to fall faster than they rise, increased volatility usually affects losses more than gains. Prior to the recent Energy Service debacle (-24.5% on the 5/4/98 purchase), no trade generated by the system had lost more than about 15% with the exception of some weeks prior to the 1987 crash. Moreover, such potential losses were relatively infrequent. However, in 1997-98, the frequency of larger negative results has been increasing.
Given the increased volatility and the consequent increased risk, I believe a change in tactics is appropriate. In particular, I now think that a stop-loss should be used on all trades. That means setting a level below the initial purchase price at which the fund will be sold regardless of what the system says or whether there will be a short-term, 0.75% penalty on the sale. Given the mechanics, that means making the calculation after the close and placing the sell order for 10:00 AM execution the next market day. The fund price will probably be different, perhaps lower, when the sale is made. Based on my research, the appropriate stop-loss level is down 6-8% from the entry price. With the process described above and the likely penalty, the actual loss will probably be a percent or two greater.
Previously, my thinking was that the purpose of stop-losses at this level was to avoid situations like the 1987 crash. Otherwise, they would be relatively infrequent, and their long-term effect on investment returns would probably be small. However, I now think that stops may be become more frequent. There is not yet enough evidence whether they will actually improve returns, but avoiding losses like the ones in Energy Service is worthwhile from a psychological standpoint. In this instance, the stop-loss sale, including the short-term penalty, would have been about 10% during the week on May 18 (when the system was signaling that Energy Service was a buy before it started its fast plunge). Presumably one moved to the money market fund and then re-entered the next week by buying Health Care because the market as a whole was not particularly weak. That fund is down less than a percent as of June 15, so in this case, the stop would have saved about 14% of the 24.5% loss. Future instances may not work as well, but it only takes a couple of losses like the current one to "beat one down" and cause the abandonment of what I still think is a good trading system despite the current poor performance.