Jim Dean
 Sage
  Posts: 3022
Joined: 9/21/2006
Location: L'ville, GA
User Profile |
This is a much discussed topic. Indepth solutions can be found here:
http://tradetight.org/forums/thread-view.asp?tid=550#m2598
Another way to handle this without approximations is to put the scan criteria into a Filter Block, with "Current+1" setting and weekly Data Periods active in addition to Daily. This is not strictly a "scan" since it is part of a strategy.
Yet another way is to write a simple OLang Indicator with the weekly rules you want in it, that returns 1 if the rules are met and 0 if they are not. Then set Data Periods so Weekly bars are adequate for your rules, and add the OL Indic as a Focus List column. Choose a List that has your entire Starting Population in it. Now, ToDo-Recalc and sort on that column. All the symbols with "1" in the column pass the test. Highlight and Ctl-C copy those symbols, and create a new custom FL with them in it ... that's equiv to what OmniScan would have produced if it could be run on weekly bars.
Yet another way is to approximate Weekly with Daily Bars. A week is approximately 5 days. So, multiply the parameters of your rules by five and SBAS (step back and squint), and use those rules for OmniScan. The problem is that a true weekly EMA(50) is not the same as a daily EMA(250) ... but its a fast and dirty solution.
|