Replies to post #109 on J.M.HURST

[JLS]

Trend1,

since you asked the question ...

No, I don't "feel" Hurst is telling me much of anything. I think it's a coin toss. Did you have that same feeling on 09/23 or on 09/17? What about the three weeks at the end of June during which you set your daily Envelope system to pretty much ignore price swings?

What I find interesting about Hurst, though, is you used both a fast and a slow channel and set the parameters so that the boundaries of each will be tested often but not exceeded often. I think that is correct.

However, in your Envelope system, which should be in competition with this method, you set the 30-min Envelope so that its boundaries are tested often, but you set the daily Envelope so that it is seldom tested. That is not correct. And it would not make a good comparison of that system to the Hurst system. They should be set similarly for a valid comparison.

This brings me to wanting to make another point. Bollinger uses a moving average also, but it uses standard deviation (as opposed to a manual fit using percentages). In your Hurst implementation, and also in your Envelope system, you use manual fit and leave it there essentially forever. Using standard deviation makes Bollinger bands adaptive to price volatility (instead of fixed), which I find to be a superior idea. And Bollinger explains that the moving average length should be set to fit the underlying price changes so that the boundaries are frequently tested but infrequently exceeded. (So most people are in error when all they do is forever use 20, the default value, for a Bollinger moving average.)

My same argument applies to modern Keltner channels. Keltner channels were modernized by Raschke to use a percentage of ATR for bandwidth, instead of a fixed value to set the bandwidth, therefore making the bandwidth adaptive, which I again think is a superior approach. And again, the period of the Keltner channel should be adjusted in the same manner as I described for Bollinger bands.