My current research passion is the interaction between decision factors and indicators in trading systems. The most interesting interactions are usually between entrance decision factors and exit factors. This is simple enough when we are looking at two-factor interactions, but many times multiple factor interactions also seem to be critical.

I've tried a couple of different ways to show three-factor interactions. My latest attempt is a quasi-animation. The basic response surface graph shows how system profitability is changed by the major interaction between the most significant entrance factor and the most significant exit factor, and the animation shows the whole surface moving as a second exit parameter is modified.

Here is a sample on youtube. It looks much better if you blow it up to full screen size.

This is a work in progress. I think it is very cool right now, but next week I'll be wondering how I ever could have shown something this crude. I like it because I think the relationships are more important to know about than the specific optimal numbers which often represent an over-fitting.

Please let me know if you have a better idea. I am kind of stuck right at the moment looking for a way to show four-factor interactions. I can see that there are four-factor interactions in the numbers, but my brain just hurts when I try to add another dimension. I would love to hear any thoughts on this as well.

In case anyone is interested, the system is a simple mean-reversion system roughly based on Larry Connors' writings. I don't think that is too important because we see significant interactions going on in all of the systems that we have tested.





Speak your mind


Resources & Links