The second project is to assess the paper of Gusta§son, “Evaluating the Longstaff-
Schwarz method for pricing of American options”. A version can be found here
Your assessment should cover:
1. That you have understood Gustaffsonís analysis of the problem and choice of algo-
rithms. Make any appropriate comments if you agree or disagree.
2. Independently implemented your own code, and compare the results of your code
with Gustaffsonís. Your own code might result from di§erent choices or the same,
but you need to test your code in a meaningful way. Hint: it is sometimes easier to
gain confidence in the results of code if you have two or more di§erent implementa-
tions. (See the solution guide for the first project for examples of how I did this to
build conÖdence in my sampling codes.)
3. Gusta§sonís figure 2.1, which appears purely qualitative, motivates his implemen-
tation details well. And he does test the error and compute confidence levels for his
algorithm. But he does not compute a version of figure 2.1 as a result of his code.
You should provide a graph of the exercise boundary as computed by your code.
4. You should be able to price the American call option and compute the exercise
boundary. This is not because anyone is curious about what that exercise boundary
is, it is because you want to see how your code performs in computing exercise
boundaries. Provide a graph shows the exercise boundary. If your code uses an
iteration, you should show a graph that illustrates the convergence of your algorithm.
The due date is October 29, 2021, which is a week later than the date in the syllabus.
This date was moved back as a result of moving back the date of the Örst project. The
TA will post details for the deadline.