Collective2

Traditional cycle analysis attempts to isolate only a few cycles. This approach, while occasionally successful, ignores the totality of all cyclical forces impacting a market. Chaos Trader identifies all the predominant cycles in a data stream, extracts them, and then does a cyclical summation. The program then builds the composite of all the important cycles into a sinusoidal-looking wave that is then projected into the future.

Thinking outside the box – The math that has gone into the program is unique and is the result of a new approach to cycle analysis. Similar to but vastly superior to Fourier Analysis, our creative approach is not available anywhere else. The technique for extracting cyclical activity, and also detrending raw data, is what allows us to produce unique analysis.

Chaos Trader methodology employs proprietary statistical techniques to obtain cyclical information from price data. Other proprietary frequency domain techniques are then employed to obtain the cycles embedded in the price data. It is our experiences that these cycles, projected into the future, have great predictive value and can be used in a trading system. This approach is generalized and accessible and avoids “over fitting” the data like more obscure approaches such as neural networks.

Chaos Trader goes through an initial de-trending filter that is determined by its specific time interval. The de-trending adds considerable computational stability to the analysis. The properties of the analysis are such that, in general, the time function being analyzed is converted into a discreet set of time dependent segments, which, by themselves, are adaptable to direct Fourier computations with variable evaluation lengths. Algorithms are applied which slide forward and back to test the validity of each transform length until minimal differences are established. Further, as these length searches are effected, another search is conducted which seeks a characteristic waveform applicable of the time segment under analysis. Once these parameters are gleaned, then the inverse of the filter is applied to the converted signal, and the lengths and waveforms obtained in the search stated. Using the inverse of the parameters in the original time domain a time variant nonlinear filter is employed on the initial function to, essentially, remove the waveforms and lengths found in the nonlinear domain.

We then use a dynamic risk management program as well. All trades are protected in some way. We use a sliding scale between -10:1 and +10:1 for leverage most of the time we are using less than 5:1. We employ a scaling in and out method so we do add to losing trades. But that is not averaging down. Scaling in and out is a risk management method that serves to reduce risk not add it the way averaging down does. It is far too simplistic to say "never add to a losing trade".

This is an aggressive strategy and not for most people. That being said we have yielded triple digit returns over the years with these methods before coming to C2. Of course our method comes with a large amount of volatility and most traders simply cannot handle that but those who can are unlikely to be disappointed. Traders are advised NOT to use additional leverage with these signals.