Quote from QQQShort:
You are right about position sizing.
Yeah, I suspected we were talking about different concepts after posting my last message. You want to pick five stocks from, let's say, the S&P 500 and run your tests against those five stocks. What criteria would you use to pick that smaller group?
Well, let's imagine a simple system where I'm using a MA crossover. I'll use the S&P500 as my base group of stocks. I want to buy, over time, 10 stocks where the 50dma has crossed over the 100dma (this isn't a strategy I use - just for illustration). I want to size the positions based on an ATR-based position sizing technique. Then I want a report on how this did over time, and I want to optimize it both backwards and using a walk-forward technique that gets rid of data-mining basis.
All of this are basic requirements to doing system testing on a portfolio in my mind.