Summer MGT 6203 MID EXAM
PART1 – THEORY
Week
... [Show More] 1
Q1) Which of the following Evaluation metric would you use to evaluate a linear
regression model with continuous output?
A. Accuracy
B. Mean Squared Error
C. ROC-AUC curve
D. Specificity
Ans: B, while others are evaluation metrics to evaluate a classification problem, we use
Mean Squared Error to evaluate a linear regression model which gives output as a
continuous variable.
Q2) In a Simple linear regression, the correlation coefficient output came out to be 0.81.
What is the percentage of the variation of the Y axis variable that is explained by the X
axis variable?
A. 81%
B. 9%
C. 66%
D. 19%
Ans: C, the variation is given by r2, where r is correlation coefficient. In this r = 0.81.
Therefore, r2 = 0.81 *0.81 = 0.656 (0.66 approx.). So, the independent variable explains 66%
of the Y%. In this case, the remaining 34% of the variation is unexplained in the simple linear
regression.
Week 2
Q3) Why do you perform dummy encoding on your categorical variables when performing
linear regression?
A. Dummy encoding creates more data, which improves the performance of linear
regression
B. Categorical data must be converted to a numerical format to be properly processed by
a linear regression model
C. Dummy encoding creates more features, which increases R-Squared.
D. Linear regression can process dummy encoded variables faster than it can process
categorical variables
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comSolution: B. Like many other models, linear regression cannot process categorical data
properly and must be converted to a numerical format.
Q4) Which is NOT a reason why we remove one category in dummy variable encoding?
A. Removing a category reduces multicollinearity between dummy variables
B. The removed category is used as a base or reference during analysis
C. Removing a category improves the inference of the model’s results
D. Removing a category reduces the time to process the data.
Solution: D, We remove a variable when dummy encoding because it reduces
multicollinearity between dummy variables, and by reducing multicollinearity you improve
the inference of your results. The account for the missing category, you use it as a base or
reference in your linear regression model. Removing a category DOES NOT improve analysis
time.
Q5) If I want to use a linear regression model to capture and approximate the elasticity
between price and market demand, which model do I create?
A. Linear-Log Model
B. Log-Linear Model
C. Log-Log Model
D. Polynomial Model
Solution: C, Elasticity is the %change in Y when X changes by %1. Log-Log models provides
that relationship.
Q6) Which is NOT true for Polynomial Model below?
y = b0 + b1*X + b2*X2
A. The polynomial model does not allow for an isolated interpretation of coefficients
B. The impact of X on y is inconsistent
C. Add b1 and b2 to find the change in y for one unit of change in X
D. b0 is the y intercept
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comSolution: C, The change in Y is different along different parts of the curve, therefore b1 + b2
does not capture those changes.
Week 3
Q7) The logit function is the log of odds function. What is the range of logit function?
a. (– Inf , Inf)
b. (0,1)
c. (0, Inf)
d. (- Inf, 0)
Ans: Self explanatory
Q8) For the following confusion matrix for an experiment:
Predicted: 0 Predicted: 1
Actual: 0 50 10
Actual: 1 5 100
What are the accuracy, sensitivity values respectively?
a. 0.91, 0.83
b. 0.67, 0.95
c. 0.91, 0.95
d. 0.67, 0.83
Ans: Accuracy = (TP + TN)/Total = (100+50)/165 = 0.91
Sensitivity = TP/(TP+FN) = 100/(100+5) = 0.95
Q9) Using a logistic regression model, an analyst is considering two models:
Model 1 with only one predicting variable A.
Model 2 with predicting variable B in addition to variable A.
The analyst notices that the sign of the estimated coefficient for A is negative in Model 1
and positive in Model 2. This is most likely because:
a. A is significant in Model two.
b. B is significant in Model two.
c. A and B are uncorrelated.
d. B is correlated with A.
Ans: Multicollinearity often leads to such behaviour between predictor and response
Week 4
Q10) Which statement is NOT true about correlation?
A. Correlation has no units
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comB. R-Squared in a simple regression is equivalent to the square of the correlation between
the independent and dependent variable
C. Correlation does not mean causation
D. A correlation of zero means there is no relationship between two variables
Solution: D, A correlation of zero means there is no linear relationship, although a nonlinear
relationship could exist.
Q11) The Difference Estimator measure what?
A. The interaction between two terms
B. The difference between the mean of two groups
C. The impact of one group on the response variable
D. The correlation between two groups
Solution: B, Lecture “Randomized Controlled Experiment and the Difference Estimator”
slide 4 states “the difference estimator … is the difference between the sample means of
the treatment and control groups.”
Q12) Which statement is NOT true for Difference-In-Difference Estimator?
A. The Difference-In-Difference Estimator can be modeled using an interaction term in a
linear regression
B. The Difference-In-Difference Estimator measures the difference in change between two
groups over time
C. The Difference-In-Difference Estimator cannot be negative.
D. A Difference-In-Difference Estimate of 0 means the event had no change on the
treatment group
Solution: C, The Difference-In-Difference Estimator can be positive or negative. A negative
DID indicates a negative change in the treatment group.
Week 5
Q13) Using the below order book please answer the following question. A trader places a market
order to buy 2100 shares of stock ABC. Assuming no delay, commission, or impact costs, what
would be the cost incurred by the trader to place the order?
Order Book (Stock ABC)
Order Type Price Quantity
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comAsk $12.50 300
Ask $12.30 800
Ask $12.00 400
Ask $11.50 200
Ask $11.20 600
Ask $10.95 200
Ask $10.90 1000
Bid $10.80 800
Bid $10.60 500
Bid $10.50 700
Bid $10.40 2100
Bid $10.00 1800
Bid $9.80 1200
Bid $9.50 600
A. $22,330
B. $23,310
C. $21,840
D. $22,890
Solution: B.
10.9 × 1000 + 10.95 × 200 + 11.2 × 600 + 11.5 × 200 + 12 × 100 = 23,310
For questions 14 and 15, please use the following information:
Given below is the market changes that happened for a stock XYZ:
Date Price Dividend Stock Split
Jan 2017 $21.25 - -
Jan 2018 $17.5 $0.03 4 for 3
Jan 2019 $19 $0.05
Jan 2020 $16 4 for 3
Q14) Assuming no other relevant events over the period, what is the arithmetic return yielded by
holding one share of XYZ from Jan 2017 to Jan 2018?
A. 10%
B. 10.5%
C. 9%
D. -10%
Solution: A.
(17.5+0.03)*(4/3)/21.25 -1 = 0.099 = 10%
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comQ15) Assuming no other relevant events over the period, what is the geometric return yielded by
holding one share of XYZ from Jan 2017 to Jan 2020?
A. 32.6%
B. 29.8%
C. 34.5%
D. 31.27%
Solution: C.
(16+0.03 + 0.05)*(4/3)*(4/3)/21.25 -1 = 0.3452 = 34.52%
Q16) John notices Caterpillar’s stock has beaten the market in September for the past five years
consecutively, so he decides to purchase the stock at the beginning of September and sell it at the
end of the month. However, at the end of September, John observes that Caterpillar’s stock
declined over the period and had a lower return compared to the S&P 500 over the period. What
is the minimum/lowest form of market efficiency that could accurately explains this situation?
A. Weak form efficiency
B. Semi-strong form efficiency
C. Strong form efficiency
D. Inefficiency
Solution: A.
Weak form efficiency is where security prices reflect all information found in past prices and volume.
Q17) An individual is out shopping with the goal to purcahse 1 baseball bat. A retail sporting
goods store sells wooden baseball bats for $50 a piece. A wholesale sporting goods store sells a set
of three of those same wooden baseball bats for $125. Which behavioural bias does the wholesale
sporting goods store attempt to exploit or benefit from?
A. Overconfidence
B. Loss aversion
C. Recency effect
D. Anchoring
Solution: D.
Anchoring is when individuals tend to act based on a single fact or figure that should have little
bearing on their decision, while ignoring more important information. In this case, individuals anchor
on the money-saving aspect of the wholesale offer. This offer is temping and convinces them to buy
more units even when they do not actually need them.
Week 6
Q18) After running a 3-factor regression model with factors Rm-RF, SMB, and HML on the
daily price returns of the Boring Company, you receive the following output:
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comWhich of the following statement(s) if any are true? (Select all that are true)
A. Based on the Boring Company’s negative coefficient value for HML, the company
would likely be classified as a value company.
B. At the p-value significance threshold level of “a = .05”, all coefficients are deemed
statistically significant.
C. Based on the Boring Company’s negative coefficient value for SMB, the company
would likely be classified as a large cap company.
D. Based on the Boring Company’s negative coefficient value for HML, the company
would likely be classified as a large cap company.
E. Based on the Boring Company’s negative coefficient value for HML, the company
would likely be classified as a growth company.
Answers: B, C, and E; (Week 6 Lesson 3 Slides 10, 11, and 12); Answer A is incorrect
because a negative coefficient for the HML factor would imply growth and a positive
coefficient would imply value. Answer B is correct because each coefficient has a p-value
less than the .05 threshold making them all statistically significant. Answer C is correct
because a positive coefficient for SMB indicates small cap and a negative coefficient implies
large cap. Answer D is incorrect because the HML factor distinguishes value and growth only
and could not give us any indication on the company size. Answer E is correct because for
the same reason Answer A is incorrect.
Q19) In recent times, one of the most popular strategies on Wall Street has been to
invest in high growth (or equivalently low value) stocks. People who jumped on the
strategies early with names like Shopify, Netflix, Tesla, and Beyond Meat have
drastically outperformed the broad market by betting on the growth factor (or
equivalently against the value factor). Famed value investors such as Warren Buffett
and David Einhorn who made their fortunes betting on the value factor prior to 2008
are now experiencing some of the worst returns in their investing careers.
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.comGiven your understanding of the long run behaviour of the value factor which of the
following statements is NOT true.
A. Although the value factor has underperformed in recent years, it is normal
for systematic risk factors such as value to experience times of cyclical
fluctuation.
B. As more investors put money into high growth stocks the high growth
premiums presently observed are more likely to be crowded out and diminish
in the long run
C. The value factor as we know it no longer exists
D. As more investors flee high value stocks (due to the presently negative factor
return premium) the future return premium for the value factor could likely
increase
Answers: C; (Week 6 Lesson 2 Slide 13 & 14); Answer A is true it is consistent with our
knowledge of long-term factor behaviour in that systemic factors such as value will go
through periods of both outperformance and underperformance due to cyclical factors.
Answer B is true and is consistent with the principals of arbitrage that as more people join in
on a profitable trade the profit will eventually disappear as competition for the assets push
prices up wiping out any excess return premium. Answer C is false and is the correct answer
choice as its recent underperformance of value is not sufficient-enough a time period or
persistence to warrant a rejection of the factor especially considering it allows for cyclical
fluctuations. Answer D is also true as the inverse case of answer B. As investors continue to
sell value stocks, eventually they will become so oversold in comparison to high growth
stocks that a positive premium will eventually likely emerge.
Q20) True or False?
The long-term existence of a positive momentum return premium represents additional
support for weak-form, semi-strong form, and strong-form market efficiency.
Answer False: (Week 5/6). The long run persistence of a positive momentum return
premium indicates that by using presently available pricing information we can expect an
abnormal return in the future. This is inconsistent (but not enough to totally disprove) all
three cases of market efficiency as they are all underpinned by the assumption that no
abnormal return can consistently be realized by using historical price information.
This study source was downloaded by 100000834091502 from CourseHero.com on 11-13-2021 19:46:34 GMT -06:00
https://www.coursehero.com/file/68220480/Midterm-Part-1-Solutionspdf/
This study resource was
shared via CourseHero.com
Powered by TCPDF (www.tcpdf.org) [Show Less]