To shed further light on that observation, I went back and updated my semiconductor supply and demand spreadsheet, separating out the periods when demand exceeded supply and charting the progress of the SOX index during those periods. Because there is a lag period before the Semiconductor Industry Association reports the data (June’s results were reported in early August) I use the subsequent monthly closing price for the SOX (in this case July) for my starting point. I also map the returns until the month after the excess demand situation no longer exists, since there is no way of telling the “last” month of one cycle until the next one has started.
During the last four cycles, owning the SOX during periods such as this one has produced returns ranging from a high of nearly 73% (2003) to a low of -10% (2001-02). Cumulatively, a long-only strategy of owning the SOX during such periods and avoiding it at all other times would have produced cumulative returns of 188%, compared to just 92% that would have been earned by a buy-and-hold strategy starting in April 1998.
What’s more, there have been relatively few periods during which this strategy would be underwater. That’s not to say that losses aren’t possible - in fact during two of the four cycles there were losses of 25% or so at some point. But those periods were relatively short term. I have taken exposure by selling put options, so if such a decline does occur I will be holding the index (in this case the SMH).
Statisticians and quantitative analysts will find much to criticize about this strategy, which is simplistic and based on a limited number of observations. Then again, many of those same analysts are the fellows that designed the complex highly leveraged derivative strategies that are blowing up right now. For me, I prefer a simple model that intuition tells me ought to work and which has, in fact, tended to work most of the time.
Disclosure: author has a short position in SMH put options at time of publication.