Hi everyone, I have just written a Matlab programme to price an American Put using LeastSquare Monte Carlo method for my dissertation, the results do converge but seems to be very under priced... here's the input:Initial stock price = 36Strick price = 40Interest rate = 0.06Time to maturity = 1 yearVolatility = 0.20Number of time steps = 50Number of paths = 100,000With LeastSquare Monte Carlo, the result converges approximately to 3.7323 , however according to Longstaff and Schwartz paper, this value should be 4.472. I have also used Binomial methods to price this Put and it agrees with the paper. Can anyone sugguest why this is happening? Since my results converge, does it mean my optimal exercise algorithm is correct? Please helpOscar

Acording to Option! the price should be around 4.487, so the Longstaff and Schwartz paper seems roughly correct. You may have a bug, or perhaps you need to increase the number of time steps...

i do not think that the problem is with the number of time steps or paths.. just have a look at the 17th and 18th page of the following study...for fx barriers 10000 paths is even enough for the price to converge....http://www.riskcenter.com.tr/Arsiv/2006 ... 159&kat=11

Yes, I have checked my programme again and I made a mistake in defining the discount factor, and the value does fluctuate around 4.48. By the way, if I want to use Antithetic Variates, all I have to do is just store the original set of random numbers and create another set which is just negative of the original number, and then carry out the calculations again right?ThanksOscar