The University of Sydney Page 1
QBUS6860
Visual Data Analytics
Weekly Assignment 9
Dr Demetris Christodoulou
Discipline of Accounting
MEAFA Research Group
http://sydney.edu.au/business/research/meafa
The University of Sydney Page 2
Weekly Assignment 9
o The option market for equity investment revises its volatility following
the disclosure of corporate earnings
o Option volatility is a measure for the degree of investor uncertainty
o When companies publish their corporate earnings, the earnings can
be surprising in terms of whether they meet, fail to meet, or exceed
financial analysts’ expectations. This is known as the ‘earnings
surprise’ and is an important form of news to the markets for revising
its prices and volatility
o The option market will revise the volatility in accordance to whether
the earnings surprise is positive (i.e. good news) or negative (i.e. bad
news), but also in accordance to how large or small is the earnings
surprise. We consider zero surprise as positive news.
The University of Sydney Page 3
Weekly Assignment 9
o So two defining factors for the revision of volatility following the
news announcements is the sign of the earnings surprise and the size
of the earnings surprise
o A third important factor that determines the revision of volatility is
the level of prior volatility. Remember that volatility is a measure of
uncertainty. If the market already felt certain about the state of the
company’s profitability (i.e. low prior volatility) then it might be
shocked more from a large earnings surprise, by comparison to when
the market already feels quite uncertain (i.e. higher prior volatility)
so it will not be surprised as much
o Finally, if the shock is considerable at the time of news release, then
the revision in uncertainty may take time to complete and there
might be a drift past the earnings release date. This is important
information as it tells us about the state of uncertainty of the option
market
The University of Sydney Page 4
Weekly Assignment 9
o The Graph Objective is “The option market reaction to earnings
release”. The graph objective must analyse the following three effects:
(1) The effect of type of news on volatility: positive or zero earnings
surprise vs. negative earnings surprise;
(2) The effect of magnitude of news: here you need some statistical
definition of magnitude for earnings surprise;
(3) The effect of prior uncertainty: you can think of prior-event
uncertainty as the volatility observed during18-20 days prior the
earnings announcements
The University of Sydney Page 5
Weekly Assignment 9
o The traditional graphing approach to visualizing the Graph
Objective and the three effects is through a collection of timeline
plots, as shown below, where Q1, Q2, Q3, Q4 indicate quartiles of
earnings surprise (i.e. a definition of news magnitude)
The University of Sydney Page 6
Weekly Assignment 9
o The traditional timeline approach has several decoding problems:
(1) Although it is easy to decode the direction of the shock effect at
the event date t=0, shown as a sharp decrease in volatility, it is
very hard to compare the relative shock effect across the many
levels of earnings surprise, i.e. we cannot see the relative decline
(2) It is difficult to compare the relative shock effect across bad news
and good news
(3) Although in the good news graph it is easier to see that the level of
the post-event volatility 20 days before is lower than the level of
the pre-event volatility 20 days after, it is difficult to say what is
happening in bad news
The University of Sydney Page 7
Weekly Assignment 9
o Notice how the traditional timeline approach applies a Data
Reduction approach
o The original dataset, option_volatility.csv, contains 6,191,574
observations (151,014 events × 41 days of volatility each over t=-
20,-19,…,0,…,19,20)
o The traditional approach reduces the data to just 328 data points:
2 graphs × 4 timelines per graph × 41 medians of volatility for
each day t for each timeline.
o Then, the traditional approach connects the median volatilities for
each day per quartile of volatility for good news and for bad news
o It is up to you to follow a Data Reduction approach or not. If you
choose a Data Reduction approach, then you are free to reduce the
dataset to whichever degree and to apply whichever statistical
summary you deem necessary
The University of Sydney Page 8
Weekly Assignment 9
o As a first step, you are required to reproduce the traditional
approach using the Stata to Tableu video that is available on
Canvas on “Data reduction to statistical summary”
o Then you must report one graph for solving the graph objective and
addressing the limitations of the traditional approach
The University of Sydney Page 9
Weekly Assignment 9
o All VDA work must be done with Tableau. Any data management
can be done with other software but you must submit your managed
data with the packaged .twbx file. You must also submit your
managed dataset in .csv form.
o You are required to submit a Word file report documenting your
data protocol, data transformations and any data reduction steps,
so that one that reads the report would be be able reproduce the
analysis. This report must be detailed and clear.
o You must also report your key findings with respect to the Graph
Objective.