Market Microstructure, COMP0049, 2023-2023
Coursework 100%
Copyright By PowCoder代写 加微信 powcoder
All questions and sub-questions are mandatory.
• Answers should be typed in where possible. In any case be aware that answers need to be
CLEARLY readable in order to be marked.
• Part of the Python code used in class to clean and pre-process LOB data can be used.
• Answers will be checked for plagiarism (also with lecture notes and slides).
• Answers will be marked for scientific quality and clarity of explanations.
• Submit ONE pdf file with all answers and screenshots of code snapshots (for question 1).
• Maximum Word limits are strict (no tolerance is applied).
• The coursework guidance folder contains LOB data to be used in question 1 and the
research paper on which part of question 3 is based on (”Measuring and Modeling Exe-
cution Cost and Risk” by R. Engle et al.).
COMP0049 1 TURN OVER
1. Empirical properties of LOBs [40 marks]
For this question please consider the LOB data attached for 5 stocks, namely AMZN,
GOOG, INTC, AAPL, MSFT. To perform the analysis and produce the plots you can
use a programming language of your choice (e.g., Python, Matlab).
a. Compute and plot the mid-price m(t) and its moving average over 10 minutes for
the ticker APPL.
b. Calculate and plot the log-returns versus time for GOOG.
c. Within the message file of the AMZN LOB data consider only the visible exe-
cutions events. Consider the trade signs of the visible executions (�t) and write
a function to compute the lagged autocorrelation C�(τ) as a function of the lag
τ = 0, . . . , 150. Briefly discuss the observed statistical properties of the order flow.
[Max 150 words]
d. What are the two possible explanations for the empirically observed properties of
the order flow? [Max 300 words]
e. For each stock, compute the relative tick size averaged over all events. Explain the
procedure and comment on the results obtained.
f. For each stock, calculate and plot the size of the first gap over time and its time
average over all events. Explain the procedure and comment on the results obtained.
g. For each stock, calculate the spread averaged over all events.
h. Produce a scatter plot of the average spread and average gap calculated for all stocks
and comment on the results obtained.
[Question 2 cont. on next page]
COMP0049 2 CONTINUED
2. Propagator models [30 Marks]
a. Provide a definition of price impact. [Max 300 words]
b. One of the measures of price impact is the following:
R(`) = 〈(�k · (xk+` − xk)〉k
where x is the mid-price (computed just before and immediately after the arrival of
an LOB event, for instance a market order arrival) and �k is the sign of the k-th trade.
Discuss the observed behaviour of the response function R(`) in the context of:
(i) the impact of individual trades (market orders, limit orders and cancellations).
(ii) the dependence of the response function R(`) (and price impact) on the order
(iii) the effect of non-symmetric liquidity provision.
c. Another relevant quantity to be used in this context is the so-called lagged sign-
return correlation, which is defined as S(`) := 〈rk+`�k〉, where r is the return and �
the trade sign (the average is computed over time). Show that with simple manipu-
lations it can be written as a function of R(`) and interpret the result.
d. Which are the main classes of so-called propagator models introduced to reproduce
and estimate price impact? Briefly discuss the mathematical setup and the main
assumptions. [Max 500 words]
e. Briefly describe how the propagator can be estimated from the empirical data. [Max
300 words]
[Question 3 cont. over page]
COMP0049 3 TURN OVER
3. Empirical cost analysis [30 marks]
a. Discuss the type of execution costs that investors face when submitting their orders
and how different types of cost affect different types of market participants. [Max
200 words]
b. Describe the different choices available to traders in the context of an execution of
multiple trades. [Max 200 words]
Consider the paper “Measuring and Modeling Execution Cost and Risk” for the following
questions. The paper examines transaction costs from an empirical point of view. To
answer the following questions you can refer to specific plots or tables included in the
c. Describe the features of the dataset used in the paper and explain the data cleaning
process. [Max 300 words]
d. Describe the two types of order placement strategies analysed and how they are used
to assess the risk and cost dimensions. [Max 300 words]
e. Discuss in what lies the novelty of the paper and the main results, focusing in partic-
ular on the identification of an efficient frontier for risk and cost. [Max 500 words]
f. What conclusions can be drawn in terms of the relationship between market condi-
tions (e.g., spread and liquidity), type of order (e.g., sizes) and transaction cost and
risk? [Max 500 words]
COMP0049 4 END OF PAPER
程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com