ECON 3394B: Market Microstructure &

High-Frequency Trading
Problem Set 1
Winter 2025
Instructor: Bo Liu
Due Date: 11:55 PM, Feb 9, 2025

• Write down your name and course number (ECON 3394B) on the first page.

• In answering each question, you must show all your steps and derivations to receive
full marks.

• There are 5 questions in total, for a total of 100 points.

• Teamwork with your classmates is encouraged, but please include the names of any
teammates on your submission to avoid any suspicion of plagiarism, everyone should
write and submit your own.

• Please submit your work on Brightspace before the due date; late submissions will not
be accepted.

1
Question 1: 20 points. Suppose that the best limit orders on the limit order book at
t = 0 are as follows:

Bid Ask
Price Size Price Size
11.00 100 12.00 200
10.50 200 12.50 100
10.00 150 13.50 250
9.00 100

and that:
at t = 1 a limit order to sell 150 shares at $11.50 is submitted
at t = 2 a limit order to buy 200 shares at $11.50 is submitted
at t = 3 a limit order to sell 500 shares at $10.50 is submitted
at t = 4 a market order to buy 300 shares is submitted
at t = 5 a limit order to buy 100 shares at $11.00 is submitted
at t = 6 a limit order to sell 150 shares at $11.50 is submitted
at t = 7 a limit order to sell 300 shares at $9.50 is submitted
at t = 8 a market order to buy 250 shares is submitted

a. Track the state of the limit order book (show it after each new order has arrived and
any transactions are triggered, for t = 1, ..., 8, in the trading screen format like the one at
t = 0) and the time, price (VWAP) and quantity of any transactions that take place. Record
the dollar bid-ask spread, that is, the difference between the lowest ask and the highest bid,
in the continuous market as it evolves from t = 0 onwards.

b. How do market orders affect liquidity? Intuitively, why?

Question 2: 20 points. Your fund is considering trading 10-year bonds issued by the
Canadian government, and you observe the following information:

• At 9:30 a.m., the lowest ask price is 102.31 and the highest bid price is 99.50.

• Five seconds later, a buy order for a block of $ 10 billion is executed at 102.76.

• At 10:30 a.m., you check the market again and see that the lowest ask price is 102.55
and the highest bid price is 100.02.

a. Compute the absolute and the relative quoted spread at 9:30 and at 10:30.

b. Compute the absolute and the relative effective ask-side spread at 9:30.

c. Compare the quoted spread with the effective ask-side spread (both in absolute and in
relative terms) at 9:30. What explains the difference between them?

d. Compute the absolute realized spread in the 9:30–10:30 interval.

2
 e. Compare the realized spread computed under d with the absolute effective spread at
9:30 computed under b. What explains the difference between them?

Question 3: 20 points. Consider Roll’s model presented in class. All the assumptions are
unchanged, but we do assume that the transaction occurring at time t occurs either at the
ask or bid price with probability λ, or at the midprice with probability 1−λ. Thus, 1−λ can
be seen as the fraction of trades that receive a price improvement or are crossed by brokers
at the midprice. Propose an estimator of the quoted bid-ask spread S for this case.

Question 4: 20 points. Consider the one-period Glosten-Milgrom model, where the secu-
rity’s true value v can be high (v H ) or low (v L ) with probability 21 each. Market makers are
competitive and risk neutral, and do not know v.
In each period, a single trader comes to the market:
• With probability 1 − π, he is a noise trader, who buys or sells one unit with probability
1
2
each.
• With probability π, he is an informed trader, who observes a signal about the security’s
true value.
– With probability ρ ∈ 12 , 1 , the signal is accurate, that is, it coincides with the
 

true value of the security.

a. Write down dealers’ expected profits when they receive both a buy and a sell order.
b. Compute the bid and ask prices set by risk-neutral competitive market makers.
c. Derive the bid-ask spread as a function of the signal’s informativeness. How does this
result compare with the case of perfectly informed insider trading? Is the market more
or less illiquid? Intuitively, why?
d. Verify whether, given the bid and ask prices derived at b, the insider is actually willing
to buy when his signal equals v H and sells when it is v L , that is, whether this strategy
yields positive expected profits in equilibrium.

Question 5: 20 points. Using your preferred data analysis software (e.g., Python, R,
MATLAB, stata..anything you want), replicate the plots shown on slides 14, 16, 18, and
20 of Topic 2 slides using the first 30 rows of the dataset Ch2 KrispyKreme data.xls or
Ch2 KrispyKreme data.dta uploaded to Brightspace.