We obtained several insights. We calculated the mean returns from an equally
weighted portfolio of the community stocks and an equally weighted portfolio of
singleton stocks. We also calculated the return standard deviations of these portfolios.
We did this month by month for 16 months. In 15 of the 16 months the mean returns
were higher for the community portfolio; the standard deviations were lower in 13 of the
16 months. The difference of means was significant for 13 of those months as well.
Hence, community detection based on news traffic leads to identifying a set of stocks
that performs vastly better than the rest.
There is much more to be done in this domain of economic metrics for the
performance of news analytics. Leinweber and Sisk (2010 and this volume, Chapter
6) have shown that there is exploitable alpha in news streams. The risk management and
credit analysis areas also offer economic metrics that may be used to validate news
analytics.
2.5 Discussion
The various techniques and metrics fall into two broad categories: supervised and
unsupervised learning methods. Supervised models use machine-learning algorithms
based on well-specified input variables. One may think of this as a generalized regression
68 Quantifying news: Alternative metrics
Figure 2.9.Phase lag analysis. The left-side shows the eight canonical graph patterns that are
derived from arrangements of the start, end, high, and low points of a time series. The right-side
shows the leads and lags of patterns of the stock series vs. the sentiment series. A positive value
means that the stock series leads the sentiment series.