Relating News Events to Changes in the Stock Market

To show the broad applicability of the Hamming distance matrix approach, we have analyzed the volume, i.e. the number of shares traded per day, of Apple Computer, Inc. The time period covered, from Aug 26, 2002 through Nov 25, 2002, was chosen arbitrarily. The data were arranged so that trading volumes of four consecutive trading days could be compared. We chose four day lots since we wanted to directly compare the results obtained here and the previously discussed stem cell transplantation data. Conveniently, the trading community considers four-day weeks to be informative in week-to-week comparisons rather than five-day weeks. Furthermore, we chose the standard error of the four consecutive volume data as the delta for the symbolization of each kinetic. The financial data were obtained through the public Internet source for financial data at http://finance.yahoo.com and the data are shown as linear spike-trains (Fig. 3.A).

When the daily data were plotted in lots of four consecutive days (Fig. 3.B), a noticeable variety of kinetics patterns emerged. The 4 day time span appears to be well suited to quantitatively capture an “emotion trend” reflecting the traders’ collective response to news. Isolating such trends from the background of normal trading may permit identifying more subtle trends such as speculation or market cornering. The Hamming distance matrix approach revealed that volume trends were similar during four 4-day trading periods, namely, Sep 3-6, Sep 19-24, Oct 7-10, and Nov 20-25 (Fig. 3.C). Two other groups of similar periods occur, Sep 25-30, Nov 4-7, and Oct 1-4, Oct 17-22 (Fig. 3.D). All other time periods showed a unique behavior pattern. The results indicate that the approach can isolate unique patterns amongst stock trading data.

Figure 3. Shown are the daily changes in trading volume of shares of Apple Computer, Inc. stock. A: An overview of volume traded during the period of August 26, 2002 to November 25, 2002. B: Overlay of all short kinetics. We classified short kinetics of 4 successive trading days. C and D: Result of sieving the original short kinetics into categories of similar curves. Each similarity group is depicted in a different color.

Per share trading volume is a response variable indicative of “public interest” in a particular security. Specifically, large trading volumes are typically seen when extraordinary news occurs. Discrimination between the types of news, i.e. good news or bad news, is reflected in the direction of the price variable. Retrospective analysis of news events shows that the kinetics of the “~+-“ similarity pattern shown in Figure 4.C relate to peek news events related to Apple Computers, Inc. Three of these correspond to the opening of a new retail store location. The remaining kinetic corresponds to a general news event, reflecting the overall dissatisfaction with the technology markets. The results suggest that the Hamming matrix approach is useful in identifying patterns in financial markets that are not readily discernable by other methods.