Hello all,I was wondering if people would share their thoughts on implementing trinomial (or binomial) trees for the two factor Hull-White model. In particular, I'd like to know how people deal with negative probabilities. In practice, are people satisified with the simple adjustment method proposed by Hull and White (calculate the probs for uncorrelated factors then apply a simple adjustment)?Using this simple approach, when the correlation is close to -1, the tree produces poor results (which has been looked at by Muck and Rudolf, Staley and Wicentowich).Any comments are much appreciated.thanx

For many cases, my prefered one would be ADI-like or any (operator splitting ) method, which avoids any bothering with additional assumptions in order to avoid negative probabilities.Some schemes can deal with cross-variations.But, the easiest one is :First, get rid of the cov(x,y) by defining a new variable z, where z=x+ay. And, after chose a coefficient such that x and z are uncorrelated.hope it helps,

Val,Thanks for the suggestion, I'll take a look into it. What does ADI stand for?

Alternating direction implicit methodsSeeThe following schemes:Peaceman RachfordYanenko,Marchuk,strang with predictor corrector schemeand many other operator splittings