as 0.60 (for Lucent), 0.51 (PALM), and 0.49 (DELL). Only six stocks evidenced negative
correlations, mostly small in magnitude. The average contemporaneous correlation is
0.188, which suggests that sentiment tracks stock returns in the high-tech sector. (I also
used full-day sentiment instead of only that till trading close and the results are almost
the same—the correlations are in fact higher, as sentiment includes reactions to trading
after the close).
Average correlations for individual stocks are weaker when one lag (0.067) or lead
(0.029) of the stock return are considered. More interesting is the average index of
sentiment for all 35 stocks. The contemporaneous correlation of this index to the equally
weighted return index is as high as 0.486. Here, cross-sectional aggregation helps in
eliminating some of the idiosyncratic noise, and makes the positive relationship between
returns and sentiment salient. This is also reflected in the strong positive correlation of
News analytics: Framework, techniques, and metrics 65Figure 2.8.Plot of stock series (upper graph) vs. sentiment series (lower graph). Correlation
between the series is high. The plot is based on messages from Yahoo! Finance and is for a single
24-hour period.