The purpose of the exercise is to build a sentiment indicator for companies in the
FTSE100 and tests its predictive ability on future company fundamentals; in other words,
do companies with more upbeat earnings calls have better fundamentals in future
periods?
⚫ Your cross-sectional analysis should be based on transcripts as of Q12018 (or any
transcript available before the end of Q22018) and look at the improvement in
fundamentals up to the subsequent 4 quarters (i.e.: in Q22018, Q32018, Q42018,
Q12019; where Q12018 is the base period). By "fundamentals" it is meant-at a
minimum- revenues and net income; you can add additional metrics if you deem
appropriate and with relevant justification. Please make sure your sample of
companies is sufficiently big and not smaller than at least 15 companies.（30 points）

⚫ In building the sentiment indicator, you should employ a lexicon based approach: you
should comment on your choice of lexicon and as part of your review of results you
may want to suggest possible alternatives for building your sentiment indicator and
why they'd be better. (30 points)

⚫ Use correlation analysis (explain which and why) to assess the relationship between
the sentiment indicator you have built against the changes in fundamentals up to T+4
where T has a quarterly frequency. While interpreting the results, pleas make sure to
comments on limitations of data and methodology and suggest potential
improvements (40 points)