Hi Mark & Everybody,
Now that I have been participating in the AIM discussions and having done some testing recently on my own AIM version, I get the feeling more and more that there are about 179 ways of operating and modifying an AIM...and stil counting. As a general remark it would appear to me that there is no way of comparing the various system unless we can think of an accepted Reference System.
I ran a VORTEX Test with an imposed down-up price structure and optimised the parameters for a maximum yield returning to the starting price after 4 cycles. This proved impossible to do as it is possible to get any yield I like with fiddling with the knobs. The fact is that at a depletion of cash one can run a little in the red and this improves the yield(or the Returns). The test criteria are not set down, so, how do we know what is BEST? I keep hearing about a 5% Minimimum Trade and a 10 % Buy/Sell Resistance. For the system I tested today a 40 % Minimum Trade provided the highest performance. Also, in order to get optimized I arrived at an aggressive Buy Policy and a conservative Sell Policy. This way I arrived at a Profit of $ 35500 for a $ 20000 investment at 50/50 cash to start.
The test results also differ with starting the AIM on a price rise than with starting it on a price fall. If you start with 20 % cash and the stock dips, the cash is gone in a flash. You would have been better off to start with 100 % cash and wait for the bottom of the dip. The question is: What is an objective AIM-test when you remove any market behavioural input information. I may have missed it but I have not seen this addressed before.
Suppose one takes an arbitrary share price history without knowing it in advance and feed it into the various AIM versions that exist. The results would not mean anything. The "operators" would want to fiddle the knobs to "get it right"....
Just like an arbitrary industrial control system will not usually be able to control an arbitrary process. The system must be tuned to the process in order to work at all.
I propose that we run an AIM Contest. We define the boundary conditins for a test and the Target Objective, and whatever guidelines we think are required are invoked. For all I care we hide the share price input and the we "operate" the AIM bij "fiddling" with the knobs. Let the best AIMer win and collect a price.
How about it?
