# Calculating the error of a Trinomial Model

I've been trying to find a formula to obtain the maximum relative error a trinomial model with n timesteps will incur given all other inputs as compared to the standard BSM model. I'm concerned mostly about american options but I'm thinking early exercise of the american options will only lead to more accurate estimates of the value of the option so I'm not worried about the differences between euro/am options. If such a formula for the maximum relative error does not exist any method to try and estimate the error would work. Thanks.

-