Well SG, I've been experimenting again. I went back and looked at the COG and the thought occurred to me that not only do we have to adjust the COG settings per pair but also per pair AND per time frame. One setting per pair just doesn't seem to hold up well.
So I took the Beast here as an example. The first thing I needed was the sampling rate. I decided to only use 4 hour and daily charts. I picked the number 365. On the daily, that's a whole year. On the 4 hour chart, that's 2 months. If that isn't a good enough sampling, then we're out of luck regardless...LOL!
So, the next thing I did was to look at variables on the settings on each time frame. What worked well and fit the price action on one time frame didn't match at all on the other. So here's what I came up with for the Beast...
Daily... 365 4 0 1.5 0.8
4 Hour... 365 4 0 1.0 0.618
And here's what they look like...daily first and then 4 hour. As usual, movements to extreme ends, even during trends, are suspect to strong reversals. Also, I've noticed that extreme angles up or down are subject to strong reversals as well.