CONTENTS

Exercise: Monte Carlo convergence rate for an ATM call

Black-Scholes parameters: S0=K=100S_0 = K = 100, T=1T = 1, r=0.05r = 0.05, σ=0.2\sigma = 0.2. Closed-form call price: \10.4506$.

Tasks

  1. Implement basic Monte Carlo pricing for the European call. For each of N{103,104,105,106}N \in \{10^3, 10^4, 10^5, 10^6\}, compute V^0\hat V_0 and the 95% CI half-width.

  2. Plot the half-width against NN on a log-log scale. Verify the slope is 1/2-1/2.

  3. Compute the absolute error V^010.4506|\hat V_0 - 10.4506| for each NN. Does the error decrease at the same rate as the CI half-width?

  4. Exit polling intuition. Compare the MC sample-size requirement with: an election poll with 1000 respondents has a margin of error of 3%\sim 3\%. How does this compare with the precision MC gives for N=1000N = 1000?