Just like stock market performance, semiconductor market performance can be judged using either fundamental or technical analysis. Fundamental analysis takes into account the underlying determinates of performance-in this case, forecasts of electronic equipment production, of semiconductor average selling prices and of semiconductor content within systems. Technical analysis ignores such fundamentals in favor of looking for trends in the numbers. In the financial world, technical analysts look for nonrandom patterns in asset prices in order to find short-term profit opportunities.

Technical analysis gets little respect in the financial world, mostly because market returns have been proven to be on a risk-adjusted random walk. Given the bad reputation, it's little wonder that semiconductor market research firms have ignored technical analysis. Going against the grain, I have spent the last couple of years learning the main tool of the technical analysis trade-regression and time-series statistical analysis.

Though I'm the first to admit that technical analysis is no better at predicting semiconductor market turning points than fundamental analysis (both are miserable), the models I have developed do provide a context for analyzing those pesky WSTS numbers. Taking into account seasonality, my time-series model expected sales of $12.5 billion in October. Statistical models have the benefit of providing confidence intervals; the interval for October was between $10.8 billion and $14.2 billion. So, actual sales of $11.7 billion were below expectations, but still above the lower bound of the confidence interval- weak but not surprisingly so.

This weakness does not bode well for 2003. The technical-analysis model expects '03 sales growth in the low single digits and possibly even negative. By contrast, most market research firms (including my own) expect growth of roughly 10 percent based on fundamental analysis. For the record, the model expects sales for November to come in at $11.9 billion with a range of $10.3 billion to $13.6 billion. All we can do now is wait and see.

Jeremey Donovan (jeremey.donovan@gartner.com) is Vice President and Chief Analyst at Gartner Dataquest.