In our next lesson we will take a closer look at the Adaptive Channel breakout system. We will analyze the trades closer and use what we learn to develop our actual system. Trend following systems have several problems 1) Open equity give back. 2) Low percentage of winning trades 4) Many trades have high risk and using reasonable stops causes the winning percentage to drop even lower sometimes into the upper 20's. Our question is how can we solve some of these problems for our adaptive channel breakout system ? Stay tune for the answers
Does anyone have any comments on the comparison of Adaptive Channel Breakout and the triple moving average crossover.
I would pick the 1.25 multiplier for the cycle. 1)The netprofit is sort of symmetrical around this number. 2)It doesn't make much difference because WHATEVER you pick will not be the "best" number going forward. You are just trying to get in the range of acceptable choices. 3) Significantly more trades here which lends statistical credence to this parameter choice. 4) Better performance on short side which is a big challenge to get. BUT, as in number 2 above, I don't think it matters that much whether you pick 1.25 or 1.4. Xephen
To solve the problem, the adaptive channel breakout system can provide a directional bias for intraday trading. Then we can increase the number of trades by getting in and out many times before it gets stopped out eventually JMHO. However, this causes more problems... how do we know which day is good for breakout trading in the dirction of the adaptive channel breakout system? Your ORB video didn't talk much about it... I would think certain bar patterns would help to set the daily dirctional bias, but I don't know how to test them correctly to show that there is an edge in those patterns, or it's just random profit, as doing ORB can easily get a profitable system on paper without any commission and slippage. About the parameter selection, I couldn't say it better than Xephen's post. The only question I have would be how to use the signal to noise ratio to eliminate the wrong cycle information?
Currently 1.25 is the best set of parameters. The only reason that I talk about 1.4 is that it was selected 10 years ago as the best set to use in the future. It has worked well over the past decade. I wanted to make a point that this method has had that long of a legacy and the results were published. This should answers any questions about how robust this methodology is. The signal to noise ratio requires bring MEM into the mix and CycleStudio. This is an addition addin to TradersStudio, which will be released when we release 2.0. In addition we also should have a neural network addin ready this spring. The price of both of these packages is to be anounced. I was trying to create this system with tools which are include in the standard TradersStudio package. In addition you could play with a version of Hilbert Transform if you want to program it from John Ehlers book and try it with other platforms. The signal to noise ratio is interesting, normally we want to see a ratio of more than 2-1 for a good cycle, often 3-1 or more. Hilbert gives us a smooth estimate of the average cycle. MEM give us spectra information , so we could also see the second and third best cycle lengths. The question is, if the dominant cycle has been 30 periods for example and now it is 50, but the difference in power between 30 and 50 is only 10% ,should we switch ?. Switching often creates problems and can reduce profits. This means that the methodology to handle changing cycle lengths is very important when using MEM, since it is more sensitive and give us more information to process. I have a few studies of this system I will publish on Elite Trader over the next few days.
Murray, In your experince does the adaptive channel breakout approach work on stock index markets if applied on a lower timeframe than daily e.g., hourly? I assume that it would not work on daily bars as these markets are predominantly mean reversing.
I have not tested it on stocks but in general stocks trade better counter trend then trend following at least for larger cap ones which is what I trade. I should test it on smaller cap ones. I am almost ready for our next step. I will be publishing a study in the next day or so showing final trade results versus maximum positive and maximum negative excursion on our basket. We will then analyze this data and see what we can learn from it.
Here is a interesting spreadsheet I created . I cut and pasted the custom report I created from the results workbook. It is a trade by trade report with three extra columns. Max Position Profit, Max Position loss, both of these are reported as positive numbers. The final column is the ratio of final trade profit/Max position profit. You can see from this report how bad the efficiency of even a good trend following system is. This is the reason that equity curve give back is the main problem of these types of systems. If you study this report you will see we had many trades in which we gave back 75% or more of our profits. Even with this, Adaptive Channel Breakout is still a great system for the ten years since I first published it.
Murray I agree with you about the low efficiency of most trend following trade systems; still people trade them. Is that because there is not really anything better or what ?