Chapter 12 Global Performance Evaluation 179
Looking back, the daily sigma of NAV was $45 million, which translated in an annual sigma of
$710 million. This figure was estimated over past daily movements in NAV. The annual sigma is
computed by multiplying the daily sigma by the square root of the number of trading days in a year
(approximately 250 trading days). The mathematical arbitrage models used give a slightly higher
theoretical figure equal to a daily sigma of $60 million, or annual sigma of $950 million. So the
annual risk was well below the target of 20%. The empirical measure was less than 10% of NAV
(710/7500) and the theoretical measure was 12.8% (950/7500). Because new arbitrage opportunities
were hard to find, it was decided to get back to the risk target on NAV by reducing the capital base
rather than expanding the arbitrage portfolio.
a. How much capital could LTCM reimburse to satisfy its 20% risk target (use the data from the
mathematical arbitrage model)?
LTCM did calculate VaR. You will base your calculation on the theoretical sigma measures (daily
sigma of $60 and annual sigma of $950). Remember that in a normal distribution, there is 5/100
chances to be below E(R) – 1.645
, where E(R) is the expected value and
is the sigma. There is
1/100 chances to be below E(R) – 2.326
and 1/1000 chances to be below E(R) – 3.10
.
b. What is the daily VaR at 5% and 1%?
c. What is the annual VaR at 5% and 1%? Interpret your results.
d. Assume that the portfolio of LTCM is made up of 10 arbitrages, which each have a daily sigma
of $19 million. They are all uncorrelated. What is the daily sigma for the fund?
e. It turns out that in periods of crisis the return on all arbitrages are perfectly correlated. Also the
expected return becomes zero (instead of 750 million). What is the new daily sigma for the fund?
What is the new annual sigma?
f. Under the assumptions of Question 5, what is the annual VaR at 1%?