All John Ehlers Indicators... - page 35

 
dmeliki:
Anyone interested in getting new book for a discount price of $40 contact me on this sites mailbox, or you can go to ebay and search for it. Thanks

what else do you have 4 sale?

 

Book for sale!

Currently I have that book only.

the book:

Cybernetic Analysis for Stocks and Futures: Cutting-Edge DSP Technology to Improve Your Trading (Wiley Trading)

 

You can see that change of optimization objective function really does not matter - does not stabilize the fluctuations of perfomance. Fluctuations of CG and 3CG are very high - those strategies are practically not tradeable. DSS, NonLinear and Sincorr fluctuate less however still too much in my opinion.

Krzysztof

 

CG, DSS, NonLinear and Sincorr

Here you have a screenshoots of fluctuation of performance for 5 John's strategies for EURUSD.

CG - center of Gravity

3CG - triple center of gravity

DSS - Double super smoother

NonLinear - based on non linear MA (see ppt presentation about non linear MAs)

Sincorr - sin corelated system - code in 2009 ppt presentation.

Y axis shows % of profitable trades. All strategies were optimized against three different objective functions with GA. Equity curve correlation, % profitable trades and (#winners-#losers)*profit. So if you see on Y value for example 0.8 in means hit 80% in window of 8 trades.

Krzysztof

Files:
cgp.jpg  91 kb
z.jpg  42 kb
p.jpg  104 kb
 
 

Update to Post #348 above:

In Rocket Science for Traders Ehlers gives a more complicated version of market mode detection:

Since the Cycle Mode exists for the smallest fraction of time and since most traders make the most money following a trend rather than a cycle, it is best to assume that the market is in a Trend Mode unless some very specific criteria are met. There are only two criteria to establish a Cycle Mode. First, a Cycle Mode exists for the period of a half-dominant cycle after the crossing of the two Sinewave Indicator lines. Second, a Cycle Mode exists if the measured phase rate of change is more than two-thirds the phase rate of change of the dominant cycle (360/Period) and is less than 1.5 times the phase rate of change of the dominant cycle.

The exception is when the prices are widely separated from the Instantaneous Trendline, it is possible for the Cycle Mode conditions to be met-but the Cycle Mode identification is clearly incorrect. I have therefore inserted another overriding rule for these cases. That rule is that if the SmoothPrice [the 4-bar WMA of the Price} is separated by more than 1.5 percent from the Instantaneous Trendline, then the correct market mode is the Trend Mode.

I will be glad if someone could code the simplified version in post #348,but this version also seems interesting.

 

SpectroMeter & SineWave Indicator(s)

I am looking different SpectroMeter & SineWave Indicator(s).

It seems TSD do not have search function and I found one from my own MT4 indicator library as SineWave and Spectrometer, but if anyone is capable to show correct threads for signal waves and spectrometers from this forum or anywhere else in the world wide web, I could be more than happy.

Regards,

Spectrum approach...

 

Laguerre strategy

A few screen shoots of optimization surfaces for Laguerre strategy. Its clear that is quite unstable and not so much profitable. It was applied to AUDJPY 1 min.

Krysztof

Files:
l1.jpg  68 kb
l2.jpg  69 kb
l3.jpg  74 kb
 

Modification required

igorad:
One more version of Sinewave is ready. Now it looks as you wish. I suppose that you try to create something similar to Better Sine Wave.

Hello igorad,

thank you for your modification.

Unfortunately, this indicator repaint`s the dot on the bar after the sinewave crossover. You can see how the bar pushes the dot by looking at the bar changes. Once the bar closes, the line is fixed.

Could you be so kind and help me/us please?

Thank you very much in advance!

 

Non Linear

John's Non Linear MA optimization surface. Upper stop loss 10 pips, lower stop

loss 50 pips. Looks really shaky . Applied to EURUSD 1min.

Files:
nl1.jpg  52 kb
Reason: