It's a dance.
Four steps over the last five years (since the beginning of 2008). Date annotations are mmyy. As before, I have used a one-year lag and a three-point moving average for the gold-silver ratio.
Last time we were here things didn't go so pleasantly.
Four steps over the last five years (since the beginning of 2008). Date annotations are mmyy. As before, I have used a one-year lag and a three-point moving average for the gold-silver ratio.
Last time we were here things didn't go so pleasantly.
You know what? You can easily do a numerical derivative or quotient of differences to make a phase plot. The plot will be closer to a phase plot.
ReplyDeleteWhy always do a delay?
I normally use a delay because the dating control on individual observations in the time series that I normally use (paleoclimate series) have a lot of uncertainty. In such series, the time delay is preferable as the size of the error bars in the first (and higher) time derivative will be larger than in the time delay method.
ReplyDeleteFor the economic time series, for which the timing of measurements is known with great accuracy, the time derivative method provides as good a reconstruction as the time delay method.
Actually the masthead image is a plot using the time-derivative method. I've never reconstructed a more beautiful state space.