Rather inspired by the current mania of trading funds or "trend following" of "trend following" funds, and also concerned about the rather volatile appearance of my system's equity curve, which I deem quite robust, I decided to do an experiment:
I programmed the system to write an output file with the daily values of the total equity while setting the starting capital to a round number such as 100,000, so as to mimic the way an index is constructed.
Then I named the instrument as "My System Index" or something of the sort. The big point value would be $ 1, with a tick value of $0.01 and a minimum 1 tick move. So I could trade it as an index futures contract so as to make some backtests about "trading the system's equity curve".
I did some tests with a simple Bollinger Band Breakout system, trading only from the long side. I set the enty channel at 2 times the standard deviation of the average close and the exit channel at -2 times the standard deviation of the average close. Then I ran some optimization tests of the average days length.
The optimal value that I found was at about 110 days. To my surprise there were no significant improvements over trading the system "full time" and the optimal value would have produced only 5 "entries" and "exits" over the course of 16 years of the backtest.
On the other hand, the test gave me some insights and the wish to explore further the best "timing" to deposit or withdraw funds into or from my system. Any thoughts?
Trading My Own System's Index or ETF
You use some function of the equity curve to select one of two post-multipliers on your position size: 0% or 100%.
Perhaps there might be other settings for the adjustable parameters of your idea, which are better still.
Six different post-multiplier values? -33%, 0%, +33%, +66%, +100%, +133% possibly?
Perhaps there might be other settings for the adjustable parameters of your idea, which are better still.
Six different post-multiplier values? -33%, 0%, +33%, +66%, +100%, +133% possibly?