- Messages
- 549
- Likes
- 2
I've recently been spending far too much time running metastock explorations and really needed to cut down the run times to make the process more efficient.
for example I recently extended my database from 300 stocks to 1500 and have explorations that can take > 3mins to run on a pentium II 266 Mhz machine.
Here is a list of the things I've done to help improve the situation:
I use daily delta updates instead of full loads when building the database from another application eg. Sharescope. You can also use Sharescope's tools to filter stocks according to fundamentals or sector if that is helpful.
I also pre-process my data to filter out historic prices down to 18 months instead of 5 years and add in the stock name for ease of use. This can reduce run times further down the chain but of course if you use a 200d m.a then you will need 200 days of data.
Load the data using Dataloader. I load the 1500 stocks into 3 directories so that no more that 600 are in one directory eg. ft350, smallcap and fledgling. You can then select by directory later on.
I have written a filter that I run on a weekly basis to sort the movers and shakers from the dodo's. Since the content of this list is not very dynamic once a week is enough and I also filter out penny stocks (sorry BLM fans) and illiquid shares too.
When running explorations I use the LOad Records option to limit the number of records to be processed and run all my reports against the movers and shakers list.
Now I have much more time for other things.
has anybody got any other ideas and what experiences do AIQ users have?
for example I recently extended my database from 300 stocks to 1500 and have explorations that can take > 3mins to run on a pentium II 266 Mhz machine.
Here is a list of the things I've done to help improve the situation:
I use daily delta updates instead of full loads when building the database from another application eg. Sharescope. You can also use Sharescope's tools to filter stocks according to fundamentals or sector if that is helpful.
I also pre-process my data to filter out historic prices down to 18 months instead of 5 years and add in the stock name for ease of use. This can reduce run times further down the chain but of course if you use a 200d m.a then you will need 200 days of data.
Load the data using Dataloader. I load the 1500 stocks into 3 directories so that no more that 600 are in one directory eg. ft350, smallcap and fledgling. You can then select by directory later on.
I have written a filter that I run on a weekly basis to sort the movers and shakers from the dodo's. Since the content of this list is not very dynamic once a week is enough and I also filter out penny stocks (sorry BLM fans) and illiquid shares too.
When running explorations I use the LOad Records option to limit the number of records to be processed and run all my reports against the movers and shakers list.
Now I have much more time for other things.
has anybody got any other ideas and what experiences do AIQ users have?
