Mark it was my understanding that curve fitting was akin to looking at a indicators signals and saying they would be better if we changed the indicators length from say 30 days to 14 days so we assume that since 14 days gives us better signals then that should be what we use going foward. Where a signal based on a theory like stocks go up when interest rates go down or three steps and a stumble were not curve fitted. Observed methods lead to curve fitting where methods based on economic theory should not. Irwin