What's Next?

Discussion in 'Automated Trading' started by jellob, Oct 7, 2008.

  1. jellob


    I have tested my system against the Dow 30 and 60+ stocks from various sectors. The initial results are pretty good. I will test the entire S&P 500 next. Assuming that all goes well, what will be the next step?

    I will studied total return, shares traded, probability of a loss greater than a constant. I do not know how to calculate the Sharpe ratio, but used the same concept of reward/risk to evaluate my returns. I am more conservative and is willing to sacrifice some profit for safety. What other risk variable can I use?
  2. Make sure that your results aren't curve-fitted or just random. How did you arrive at values for the variables in your system? If it was by optimization (and not forward-walk optimization) then there's an extremely high chance your system will lose money in reality.
  3. jellob


    I am not too worried about optimization. When I change a parameter I will re-test the entire basket and look for patterns. For example, if I want to reduce the number of points in a MA, I will pick 60m, 45m, 30m .... If the aggregate improves, P&L(30) > P&L(45) > P&L(60), then I know that it is legitimate.

    My only concern is that I have only one year of data -- that it is not representative.