I think the lack of response is because there are many ways to do this and because you have to use non-direct methods of doing so. Perhaps you could narrow in on your question to make it easier to respond to, such as how you intended to store the values.
--- In [email protected], "ta" <tagro...@...> wrote: > > Jeff, thanks for your comments. I'll give a try. However, I am somewhat > surprised that I did not receive more comments. I thought this issue was > something others had grappled with. > > > > From: [email protected] [mailto:[email protected]] On Behalf > Of jeffro861 > Sent: Sunday, September 13, 2009 6:46 PM > To: [email protected] > Subject: [amibroker] Re: Best Approach to a Multipass Problem > > > > > > The quickest way I know would be be to use "BATMAN" to run a batch > exploration: > 1. find beta and alpha. run first exploration and use Addsummaryrows() to > find min and max for each factor. save table (1 pass) > 2. load table ,normalize and save (could do this a few different ways) > > --- In [email protected] <mailto:amibroker%40yahoogroups.com> , "ta" > <tagroups@> wrote: > > > > I am calculating certain characteristics for every stock than I have in my > > EOD database (approximately 3000). For the sake of clarity let's use alpha > > and Beta as characteristics. I want to : > > > > 1. calculate the min and max value of alpha & beta in my data base. > > > > 2. Then normalize the values to values between 0 and 1 > > > > 3. Then calculate the sum of normalized alpha and beta and store them > > in atc file to be used in backtests > > > > > > > > I was thinking of using the following procedure: > > > > 1. Calculate alpha and beta and store the values for each stock in an > > atc file (first pass thu database) > > > > 2. Run thru the atc files and find the min and max values for alpha > > and beta and store is in one new atc file (2nd pass thru database) > > > > 3. Normalize alpha & beta for each stock using the stock's atc file > > and minmax atc file that I created during the previous step (3rd pass thru > > database) > > > > > > > > > > > > Is there a better/faster approach? Any help or suggestions is much > > appreciated. TIA > > >
