Log in

Performance limits of the Brain Trend system

  • Public
By JohnLast 3149 days ago Comments (5)

I think that the system that is available here has reached the practical limits of its performance. I made a lot of tests using number of different filters. The last thing I tried is to use adaptive laguerre filter instead of jurik filtering. Well it look good but it is not better. Why? I think that because I reached the practical limits of improving the system. With the actial type of smoothing it is a  balance between a very light version (necessary for optimization and hypoyhesis testing) and optimum smoothing.

What is the bottom line.

This is a very good system when the market is making good directionnal movement. In fact this is the best think I know in those kind of market conditions.

The only way to go futher is not to adapt the system to the market but the opposit to look the market where the system is going to perform. That would allow me to go beyond the actual limits. 

That means that I would look a market with good directionnal movements. It is not necessarily to have a trend. I am much more interested in the structure. For example the optimum performance is reached when there is a low fractal dimension at two key levels 15 m. and 30 m. 

That means that the system will perform where there is globally a low fractal dimension. When  the fractal dimension is law that the movement is prone to sharp unidirectionnal movements with sharp reversals as V tops and V bottoms (black noise reversals). The system is built in order to detect the signal from the noise. That is why I am sometimes amazed that it perform well even in conditions that are beyong its design and operational limits.

This system is the very opposite from the CHF system. The CHF system is designed to profit from antiperistent movements. The Brain trend is designed to perform in persistent movement. 

And her I do not talk about the general market direction as the trend (as it is defined by the standard technical analysis), I talk about the price microstructure, that is defined by the fractal dimension.





  • JohnLast 3148 days ago

    In fact the design of this system is to remain robust. When it works it work with almost all parameters of the lenght of the digital filter.  And it works when we have the right market conditions, we need a lot ot volatility because it is a volatility break - out system. We need also persistent movements.

    And those conditions come and go in a cyclic way on the market, that means that the key for me is to predict when the conditions are right for the system.

    And we have very clear volatility cycles on the market.

  • JohnLast 3148 days ago

    I tried almost all available to me combinations of digital filters in order to improve the system. However there are some limits, that are structurally limitated. In fact as we need to have a break - out from a certain volatility treshhold it it possible to try to adapt the volatility treshhold too. For that reason I could add this as a parameter. Once I had an idea to have different parameters for every different time of the day.

    In fact the volatility is changing during the day. As a general rule the volatility increases during the session after int decreases and the volatility is getting bigger at the opening of the US markets. And after that there is a decrease too. In order to beat the spread with that system ;) I think futher improvements are necessary.

    And this volatility profile is statistic. Do you think is there is a paractical way to change the volatility tresholds during the day depending of what hour of the day we are. For example to the basic coefficient of volatility treshold can be added additional tresholds depending of the time of the day.

    Look here I add some charts for the volatility of the Fx pair Eur/Usd.


    От 04 октомври 2011


    От 04 октомври 2011
  • JohnLast 3148 days ago

    Look at the last chart. Here I can tell you that the most profitable periods of the Brain Trend system is when the slope of the volatility is rising and the loosing periods are when the volatility slope is falling. 

    I think that information may be usefull.

  • CamaRon 3147 days ago

    This is so cool. Thanx for diggin' it up!