Quote from brettman9:
Many well-articulated ideas. Thanks for sharing.
Years ago, I reached similar conclusions to yours about designing systems from simple indicator readings. In fact, I spent about 4 years working at systems design.
I used to skip sleep between 2-4 nights a week (was bartending for income, had no capital to speak of). Taught myself to program, and started trying every recipe in the book for the first year or two.
Then I reached the point where I realized that all the indicators were just different ways of summarizing and displaying the same underlying data set: the price/time/volume trading data. The computer certainly saw it that way.
There was no added information.
So I started trying to teach the computer how to read patterns and do fuzzy logic functions to discriminate in discrete terms between trending moves and consolidations, and to parse between types of consolidations that evolved in a manner that contained the essential qualities of previous consolidations in the same market that had bullish/bearish outcomes.
For a while, I thought I was getting somewhere b/c I had managed to get a firmer and firmer sense of what I wanted the computer to be able to spot.
And that's when I started to realize an enormously valuable side-effect in the process of spending literally thousands of hours trying to teach a computer to be a good discretionary trader and find an edge in a market.
I never succeeded at the task of creating the system. But I have become a profitable discretionary trader. It took a few more years, and it helped a lot when I got into a trading office in Chicago and had the right type of environment and some important influences.
I know this is a bit off your topic. But I just wanted to point out that the process itself can be the source of an edge.
Anyways, good thread.
