QA13 - Simulation and Bootstrapping

Question 1

Describe ways to reduce Monte Carlo sampling error

Answer 1

• increase number of samples
• use antithetic random variables constructed to have negative correlation with variables used in simulation
• use control variables which have zero mean but negatively correlated with error

Question 2

Describe the bootstrapping method and its advantage over Monte Carlo simulation

Answer 2

Sampling with replacement from past observed data, less computationally expensive than monte carlo

Question 3

Describe pseudo-random number generation

Answer 3

Computers do not produce random numbers, but they run off a function using a seed that is very hard to predict

Question 4

Describe situations where bootstrapping is ineffective

Answer 4

Can only generate already observed values so can not predict shock loss unless in the data set and not robust to situation changes

Question 5

Describe the disadvantages of the simulation approach to financial problem solving

Answer 5

If market / economic situation shifts to a situation which has not been experienced in the data before, simulations will perform poorly