May 23, 2012
The Ichimoku Kinko script was improved to such an extent that it has become an enviable script with desirable long-term performance results. Over its almost 6-year test, it achieved a 47% CAGR while accumulating shares and cash in its account. Its average hit rate improved from some 35% to over 70%. 40 of the 43 stocks in the portfolio showed higher performance results. Of the 16 stocks showing losses in the original test, only 3 remained with total losses representing less than 1% of the generated portfolio profits. A very small price to pay for the added performance.
But still, a question remained: is it worthwhile to continue to improve the Ichimoku script?
Most probably, in the end, nothing of the original script will remain. Already to achieve the higher performance level the script has been greatly altered. Isn't that like designing a new script?
Striping the Ichimoku Kinko of all its trading procedures and replacing them with better-performing processes should be viewed as a “new” script deserving its own name? I might as well design a totally new script and migrate some of the enhanced trading processes as I will also want to test at the same time some new procedures. In a way it is funny, to improve the Ichimoku script it was first necessary to eliminate what was holding it back as if the script was shooting itself in the foot so to speak. And by eliminating its trading procedures, you were gradually eliminating the trading philosophy behind its design.
What was wrong with the Ichimoku Kinko script in the first place? Well, I think that having been developed in a pre-computer era, it had to be simple and in those days, they did not have the tools to backtest properly any of the trading procedures over long periods of time. Using the mid-price channels as a moving averages cross-over system is still only a slight variation on the theme. The original script had bad entry points, bad exit points, and ineffective bet sizing. Its stop and reverse on the mid-channel crossovers is poorly designed as it can trap unintentionally long and short positions for the long haul. And it is the long-trapped positions that are responsible for most, if not all, the profits generated and not the trading itself. Its other weaknesses are that it is unidimensional, wastes time and trading opportunities.
As was shown in On Optimal Portfolio IX, and in On Seeking Alpha Part III, the number of profitable trades counts. And adding shorting could just be another way to increase that number; it makes the search worthwhile. One thing is sure, it was not found in the Ichimoku Kinko Yho chartscript.
My objective was to find and modify an existing well-known script with some shorting abilities and see how I could integrate that with my own trading procedures in an effort to increase the short to mid-term trading functions; thereby generating more profits to provide more funds for the underlying accumulation processes. Adding a shorting ability would have also implied that at times the accumulating inventory would be liquidated and would require designing new functions to restore the accumulative stance. It has not been demonstrated that liquidating the inventory, under my methodology, to make a profit from short positions is worth the endeavor long term. I suspect that short term it might improve performance somewhat but as time advances, your inventory grows bigger and you start experiencing increasing opportunity costs. Like anything else, it has to be tested at the portfolio level and over the long haul.
What you seek in a trading program is its ability to generate some alpha. Not just on a few stocks here and there; but for the majority of the stocks in your selected portfolio. Even in its current state, the improved version of the Ichimoku script has demonstrated just that.
(click to enlarge)
I will have to leave the Ichimoku improvement process as is and start a new program where I will not be constrained by old market folklore. As the above chart shows, the trading procedure improvements brought to the Ichimoku script were noticeable between the two programs which started with the same initial test conditions over the same stock selection. It was only the set of trading procedures that made the difference since ΔP? was the same for both tests.
Published ... May 23, 2012, © Guy R. Fleury. All rights reserved.