Recently our company is trying to strip out the prepayment cash flow from mortgage servicing. When I'm trying to quantify the difference between actually fv and predicted fv, there is always a huge percentage error for option arm tranche(20%~50%, when the conventional fix , arm tranche are usually below 5%). By the way, after strip out pp cf, the neg am tranche for servicing looks good. Is it because the recent credit crunch? Does anybody know about this? Thx.P.S. My predicted fv is a function of duration+convexity+vega+partials+spread. My actual fv is using monte carlo simulation.

Any idea? Or if you have experience the same kind of situation, just throw some light.