I've used REGRESSION CHANNELS (RC) in other trading software and by comparison the NINJA RC appear to be much too narrow.
I assume that price is Normally distributed so if your RC is built using Close price, you'd think that 95% of Closing prices in a swing would be inside two standard deviations. As far as I can tell after drawing a Ninja RC set to 2 Std Dev and counting bars outside the channel . . . I typically find that over 20% of bars Close outside the two standard deviations. So what Ninja calls "2 standard deviations" is more like 1.5 (? just a guess).
Now you might think . . . "well, but some swings are more volatile than others" . . . and that's true. But as I understand it, the linear regression formula used to construct the center line would take increased volatility into account.
So . . . .
1) Any general thoughts welcome. Or if my understanding of Regression Channels is wrong, school me.
2) Is Ninja using something other than a Normal distribution?
3) Does anyone know of a RC (third party) that does appear to include 95% of Closing prices?
Thanks again
John
Comment