The cycle model does not have a learning algorithm like our standard model
so shocks are taken into account by recalculating the cycle parameters. It is
impossible to know a priori what effect new parameters will have on the predicted
trend. As I state on the forecast page, this model is experimental. So far, I
find that it is very useful in predicting the market direction. I plan on keeping
it for another month and I may then replace it with something else that I have
been working on that may do a better job.