Hello!I'm trying to implement LCG Rogers' MC method for pricing american options for a college project.The option I'm considering is a simple american put.This is the first example featured in the paper...While the prices I'm getting aren't totally nonsensical, they are significantly less accurate than those reported in the paper.The author talks about using Richardson extrapolation in getting the prices. Can anyone direct me to where I might find out what he means? How does one use this extrapolation to improve the price estimates?Also, if anyone has implemented this algorithm themselves, I would love to see the output they got- especially values of the multiplier lambda...Thanks.

QuoteOriginally posted by: bigmouthWhile the prices I'm getting aren't totally nonsensical, they are significantly less accurate than those reported in the papertwo possibilities of many:(i) your code is wrong(ii) (and much less likely) he faked his answers

I'm sure (ii) isn't a possibility.It is however quite possible that I've made a mistake in my code, or misunderstood the paper in some way. Or that the extrapolation step improves the results significantly.That's why I was hoping somebody might have some further results to compare with mine.

QuoteThe author talks about using Richardson extrapolation in getting the prices. Can anyone direct me to where I might find out what he means? How does one use this extrapolation to improve the price estimates?Does the author justify mathematically that Richardson extrapolation is applicable to the current problem? I have used RE in non-MC appplications and erroneous results are gotten if the function is not smooth. Edit: the funny thing with RE is the more accurate the solution becomes (theory) the more round-off errors you get.