|
PureBytes Links
Trading Reference Links
|
It might not take as long as you think if you run your optimization intelligently. For example, you don't need to optimize in small steps. If you are optimizing a parameter between, say, 10 to 100, do it in steps of 10 or 20, not 1 or 2. How many variables are you optimizing? If more than 3, you are probably using too many degrees of freedom anyway. Limit your no. of optimized variables to a small number (2 or 3) and try optimizing in as large a step as possible to still give you reasonable expectation of getting you your desired values. If your database is set to local, AB is very fast. You might be surprised at how fast you can do an optimization on your entire universe of stocks. You'll wind up with a system that may not be optimum for each stock, but it'll give you an average set of parameter values for your universe. Al Venosa
>From: "wingnut_1944"
>Reply-To: amibroker@xxxxxxxxxxxxxxx
>To: amibroker@xxxxxxxxxxxxxxx
>Subject: [amibroker] Re: Having trouble with concept of Exploration
>Date: Tue, 02 Jul 2002 19:25:14 -0000
>
>I know exactly what you mean!
>
>BTW. I have this "formula" that I've created and haven't optimized
>it yet. Don't I need to optimize it against all the stocks (might a
>take a week of run time) in the universe first? I know that can't be
>the right answer (I think) so how should I choose the value of the
>unoptimized variables to test against the universe?
>
>
>
>--- In amibroker@xxxx, "Richard Alford" wrote:
> > There are a lot of trees - I confess to generally having to hit the
>scan/explore/backtest/optimize buttons before I finally get what I
>want - but I see that as a personal problem .
> >
>
>
Send and receive Hotmail on your mobile device: Click Here
|