Monthly Discussion



As the New DOW Becomes of Age


On November 1, 1999 the Dow Jones underwent major restructuring. Microsoft, Intel, Home Depot, and SBC Communications replaced chevron, Goodyear, Sears, and Union Carbide respectively. Reorganizations of such magnitude are rare, and this one was even more special because it resulted in the introduction of an important technology bias in the Dow stocks. It is for this reason that I often referred to this revamp as a possible 杜utation of the Dow into a different 都pecies.

Experimental evidence for a species mutation is any observation of a major deviation from natural growth curves (S-curves) that the species may be following. For the DOW the daily share volume and dollar value have been growing along such S-curves, see Exhibit 3, and we indeed observe that following November 1, 1999 the data seem at odds with the S-curve (purple line) that the dollar value had been following up to that time.


Exhibit 3. The data (black line) have been smoothed by a 10-day moving average to eliminate extreme spikes). A sharp change in the trend begins on November 1, 1999 (red line) when the new DOW came into existence.

But by now, the new DOW has already been in existence for more that 1200 days, and this provides us with sufficient data to try and analyze this new species. In Exhibit 4 we see an expanded view of the graph from Exhibit 3 but beginning only on November 1, 1999.

The data are amenable to an S-curve description but it is a downward-pointing S-curve! There are fluctuations but indisputably the red line in Exhibit 4 does better justice to the data than the purple line in Exhibit 3 over the same period of time. Similarly with the share volume, see Exhibit 5.




Exhibit 4. Dollar value data and S-curve fit for the period 1-Nov-1999 to today. Once again the data have been smoothed by a 10-day moving average. It appears that following some resettlement downward, the dollar value has finally stabilized.


Exhibit 5. Dollar value data and S-curve fit for the period 1-Nov-1999 to today. Once again the data have been smoothed by a 10-day moving average. The general trend is rather flat. The red curve is an S-curve but it represents a very small step downward.


An interesting observation in Exhibit 5 is that the evolution of the share volume is practically constant. I remind the reader that this is the true volume in the sense that corrections have been made for all splits. The fact that it is so flat is not surprising. It simply indicates that the amount of real company pieces that exchange hands every day is rather constant, and that all the market movement is linked to movement in prices.

We can now forecast the evolution of the new DOW by taking the ratio of the two curves from Exhibits 4 and 5. The answer is the blue line in Exhibit 6.



Exhibit 6. Data on the weighted DJIA (yellow line) and the forecast (blue line) as described in the text. The actual DJIA (black line) and long-term forecast (red line) are also shown for the sake of comparison.


The future of the new DOW is forecasted as a horizontal line. This is in rough agreement with the long-term forecast of Exhibit 1, also shown in Exhibit 6 as a red line. The 700-hundred point difference between the two forecasts is not due to the mutation in question but rather to the fact that the weighted DJIA, that I am forecasting, has been lying significantly below the DJIA for some time now, consequence of investors preference for small-price stocks. But this is somewhat of a fad. We have seen in the past investors popularity swing to higher-price stocks.

Treating the DOW as a species born on November 1, 1999 predicts a flat evolution for the future, if at some 700 points higher than the today痴 levels. The obvious question is: For how long is this flatness going to last?

I have a rule of thumb concerning the validity of quantitative forecasts based on historical data: Never project a forecast further in the future than the length to the historical window on which it is based. In other words, we cannot count on the DOW to remain flat for more than another couple of years.