Question 1 (7 points)

Please explain the idea of statistical arbitrage, and describe an example of trading strategy using
this idea. (Hint: you could use the example of Siamese twin stocks such as Royal Dutch and
Shell).

Statistical arbitrage strategies are based on relationships between security prices that tend to be
true on average.
For example, Royal Dutch and Shell are two companies that effectively run the business as if
they were a single company, but there are still two different stocks, one traded in the UK and
one in the Netherlands. The two stocks pay the same dividends, so they would be expected to
have the same price. In practice, it isn’t the case, each stock is influenced by its own market.
However, the ratio between the two prices tends to remain close to one, and when one of the two
stocks is much more expensive than the other, it is very likely that it will go down in price, closer
to parity between the two prices.
Trading strategy: when the difference is high, buy the cheaper stock and short sell the more
expensive one, and collect the difference in prices. Then, when prices are close to each other,
close the positions, which will cost almost nothing because the two prices are almost the same.
This is not an arbitrage because there is no 100% guarantee that the prices will return to parity,
and we also don’t know how long it will take. So there may be some risk.

Question 2 (9 points)

a. Briefly explain the difference between the first pass and second pass regression used in tests of
the CAPM.

First pass: we look at one stock at a time over many time periods (so it is a time-series
regression), in order to estimate the parameters that are needed for the second pass regression
(beta, exp. return, std dev. of the specific risk)
Second pass: we include many stocks (so it is a cross-sectional regression) in order to
investigate the relationship between beta and expected return, and see if the relationship
estimated from the data is consistent with the CAPM theory.

b. Take the second pass regression (recall that b^i is the estimate of stock i’s beta.)

If the CAPM theory is correct, what would you expect the coefficients 0, 1, 2, 3 to be equal
to? (of course, each of them may have a different value.)

0 = 0, 1 = E[rM – rf], 2 = 0, 3 = 0
c. It has been found that low beta stocks tend to have higher returns than predicted by the CAPM.
How would you expect this to show up in the estimated regression coefficients? In other words,
if the finding that low beta stocks have higher returns than predicted by the CAPM is true, how
should the estimated coefficients differ from the values in part b of this question? (Hint: a graph
may help you answer this question, but is not required.)
0 > 0, 1 < E[rM – rf]

Question 3 (9 points)

A mutual fund with a beta of 0.8 has an expected rate of return of 11%. The risk free rate is 3%
and you expect the market return to be 11%.

a. What is the fund’s alpha? Should you necessarily invest in this fund, or does the answer
depend on your situation? For example, your conclusion might be different depending on
whether you are planning to invest all of your money in the fund, or whether you are considering
adding the fund to an existing portfolio.

Alpha = 11% - CAPM formula = 11 – [3 + 0.8 (11 – 3)] = 1.6%

A positive alpha tells you that you should add the fund if you already hold a diversified portfolio.
If you are considering investing all of your money in the fund, it is not the relevant measure.

b. What passive portfolio combining a market index ETF and T bills would have the same beta
as the mutual fund? (Describe the composition of that portfolio in percentages.) What is the
difference between the expected return of the mutual fund and the expected return of this passive
portfolio? Does this portfolio have the same standard deviation as the mutual fund?

80% in ETF, 20% in T bills. Exp. return of passive portfolio = 9.4%. Difference in exp. Return =
1.6%.
The standard deviations are not the same, because the mutual fund might have some specific
risk.

Question 4 (18 points)

2
a. If the CAPM is valid and the riskless rate is 5%, is the following situation possible? Explain
your answer.

Stock Exp return Std dev

A 10% 25%
B 12% 20%

Yes. The CAPM does not imply anything about the relationship between the exp. return and
the std dev. of individual stocks.

b. If the CAPM is valid and the riskless rate is 5%, is the following situation possible?
Explain your answer.

Stock Exp return Std dev

A 10% 25%
B 12% 20%
Market 10% 16%

Sharpe ratios:
A: (10 – 5) / 25 = 0.2
B: (12 – 5) / 20 = 0.35
Market: (10 – 5) / 16 = 0.3125

Impossible, because according to the CAPM, the Market should have the highest Sharpe
ratio.

c. If the CAPM is valid, is the following situation possible? Explain your answer.

Stock Exp return Beta

A 9% 1
B 14% 1.5
C 3% 0

E[rC] = 3% => rf = 3%
E[rA] = 9% => E[rM] = 9%
Then according to the CAPM, E[r B] = 12%
Impossible.

d. If the CAPM is valid and the risk free rate is 2%, is the following situation possible?
Explain your answer.

Exp return Correlation with market Std dev. of return

Market 8% 1 16%
Stock A 11% 0.75 32%

A = cov(rA,rM) / M2 = AM A / M = 0.75 (32 / 16) = 1.5

3
Then according to the CAPM, E[r A] = 2 + 1.5 (8 – 2) = 11%
Possible

e. If the CAPM is valid, the risk free rate is 2% and the correlation between Pelléas and
Mélisande is zero, is the following situation possible? Explain your answer.

Exp return Std dev

Market portfolio 8% 16%
Pelléas hedge fund 12% 24%
Mélisande hedge fund 13% 32%

(Hint: consider an equally weighted portfolio of Pelléas and Mélisande).

Exp return of equally weighted portfolio = 12.5%

Variance = 0.52(0.24)2 + 0.52(0.32)2 =
Std dev = square root of variance =
Sharpe ratio = (12.5 – 2) /
Sharpe ratio of market = (8 – 2) / 16 =
Impossible because, according to the CAPM, the market should have the highest Sharpe ratio.

Question 5 (10 points)

The riskfree rate is 2%, and the market expected return and standard deviation are 8% and 16%
respectively. You have obtained the following data on two stocks:

Exp return Beta (ei)

Stock A 13% 1.4 20%
Stock B 15.7% 1.8 32%
(ei) is the standard deviation of the residual in the CAPM regression (the regression that is used,
for example, in the estimation of alpha.)

a. Which one of these two stocks would you want to add to your portfolio if you are currently
100% invested in the market portfolio? (You can only choose one of the two stocks.)

Alpha of stock A: 2.6%

Alpha of stock B: 2.9%
Both stocks have positive alpha so you should look at the information ratio
IR of A: 0.13
IR of B: 0.09
Pick A

b. By how much does the Sharpe ratio of your portfolio increase after adding the stock you chose
in part a?

SR of market: (8 – 2) / 16 = 0.375
Best possible SR = [(SR of market)^2 + (IR of A)^2 ]^(1/2) = 0.3969

4
Difference = 0.0219

Question 6 (12 points)

The price of the Kramerica Industries stock is currently $125. Call and put options on the stock
with an exercise price of $120 and a maturity of one year have prices equal to $16 and $12
respectively. The riskless interest rate is 1% per year. (You can invest and borrow at the same
1% interest rate.)

a. What is the price today, and the payoff one year from now, of a portfolio including one share
of the stock and one put option? (You need to consider two cases, depending on whether the
price one year from now is more or less than $120. In each case, the payoff is either a number, or
a simple function of the stock price.)

Price = 125 + 12 = $137

Put payoff: 120 – S if S<120, 0 otherwise.

Stock payoff: S
Portfolio payoff: 120 if S<120, S if S>120

b. Can you find an investment strategy involving the purchase of one call option and an
investment at the riskless rate that will provide the same payoff (in one year) as the portfolio in
part a? Provide the exact amount that needs to be invested at the riskless rate. What is the cost
(today) of this strategy?

Buy one call and invest $X at riskless rate.

Call payoff: 0 if S<120, S – 120 otherwise
Riskless investment payoff: X(1.01)
Total payoff: X(1.01) if S<120, S – 120 + X(1.01) if S>120.
So if X(1.01) = 120  X = 120 / 1.01 = $118.81

Cost = 16 + 118.81 = $134.81

c. Is an arbitrage opportunity available? If yes, please describe the strategy in detail. What is the
profit (per share of stock involved in the arbitrage)?

There is an arbitrage because the strategies in a and b have the same payoff in one year, but
different values today (137 vs. 134.81). So you should buy the cheaper and sell the more
expensive one. Specifically:
Buy one call
Invest $118.81 at the riskless rate.
Sell one share of the stock
Sell one put.
Immediate profit = $2.19 (137 – 134.81)
In one year, cash flow = 0 (no risk).

Question 7 (12 points)

5
The Satriale stock is currently trading for $232 per share. A forward contract on the stock is
available, with a maturity of one year. The interest rate is 7% per year.

a. Determine the forward price.

Replicating strategy:
Borrow $232
Buy share
In one year, reimburse loan => CF = - 232 * 1.07 = $248.24
So the forward price should be $248.24

b. Assume that the forward price equals $250 (rather than the answer you determined in part a).
Is arbitrage available? If it is, describe the arbitrage strategy as precisely as possible.

Do the replicating strategy and sell forward at the same time.

In more detail: at t=0, borrow $232 and buy 1 share, and sign a forward to sell 1 share in year.
At t = 1 year: deliver the share, receive $250, pay $248.24 to repay the loan
Profit: $1.76

Question 8 (13 points)

The risk free rate is 4%. Assume that returns on individual stocks are generated by the following
two factor model:

Ri = constanti + 1i F1 + 2i F2

Security 1 2 Exp. return

1 1 1.5 20%
2 2 2 20%
3 1 1 10%
4 1.5 1 10%

a. Construct a portfolio containing (long or short) securities 1 and 2, with a return that does not
depend on factor F1. Compute the expected return and factor sensitivities 1, 2 for this portfolio.

x1 (1) + x2 (2) = x1 (1) + (1-x1) (2) =0 => x1 = 2, x2 = -1

Exp. return = 20%, 1 = 0, 2 = 2 (1.5) – 1 (2) =1

b. Construct a portfolio containing (long or short) securities 3 and 4, with a return that does not
depend on factor F1. Compute the expected return and factor sensitivities 1, 2 for this portfolio.

x3 (1) + x4 (1.5) = x3 (1) + (1-x3) (1.5) =0 => x3 = 3, x4 = -2

Exp. return = 10%, 1 = 0, 2 = 3 (1) – 2 (1) =1

6
c. Is there an arbitrage opportunity? If the answer is positive, describe the arbitrage strategy as
precisely as possible. (Hint: it is not necessary to consider the riskless asset to answer this
question.)

The portfolios in part a and b have the same factor sensitivities (to both factors), but different
expected returns. So by short-selling the portfolio in part b and buying the portfolio in part a,
you pick up the difference in expected returns and face no risk.
For example, you could buy $200 of stock A, short-sell $100 of stock B, short-sell $300 of stock
3 and buy $200 of stock 4. The profit would be $10.

Question 9 (10 points)

What are some risks and difficulties of short selling? (We discussed many of these in class. You
are not expected to remember all of them. Please be concise.)

The stock may go up instead of down, leading to margin calls and losses.
If you receive a margin call and cannot post additional margin, your position will be terminated
at a loss (even though you may be right in the long term).
There is no theoretical limit to the amount that can be lost.
On average, stocks tend to go up, so the odds are not in your favor.
The stock may be costly or even impossible to borrow.
There is a risk that the stock you borrowed may be recalled by the lender.
There is the risk of a short squeeze: if too many short sellers want to end the short sale at the
same time, they need to buy, which causes the price to go up, leading to losses.
Short selling is unpopular, which causes restrictions to be imposed on short selling from time to
time (especially in times of crises).
Companies sometimes fight short sellers by suing them.