defghca
150 posts
msg #26959
- Ignore defghca |
6/20/2003 7:20:49 AM
i see a lot of filters here that do not live up to their creators expectations or claims.
filters will hopefully guide your entry points into a stock, further analysis is required. to be complete it should be completed with money management rules and most important exit rules. all of these make a complete system . what we have now is entry filters apart from muddy's LBB system wich is complete
also we should devise a reporting system to grade filters.parameters could be# win,#losses, win/loss ratio,$/bar held, sharpe ratio , maximum drawdown ect
backtesting using SF " date offset" is primitive at best. i suugest the use of the following free sites for real backtesting capabilities for your filters
not to many filters will holdup to their claims after going through these ringners
http://www.wealth-lab.com/
http://tbots.com/
http://www.lathuy.com/
|
janeiro
25 posts
msg #26973
- Ignore janeiro |
6/21/2003 2:14:33 PM
defghca....
let me thank u for the address that u provided in your post...Interesting site...
THX for sharing
Nymbus
|
ig0r
51 posts
msg #26974
- Ignore ig0r |
6/21/2003 2:56:54 PM
I agree, detailed performance of a filter should be given, or it will not be taken seriously
|
knowsenough
54 posts
msg #26976
- Ignore knowsenough |
6/21/2003 5:19:06 PM
defghca....
Ditto that previous post for those backtesting sites....I started scripting scans here back in March '03 and kept beating on them to shrink the number of hits/scans till I got them down to (8)scans w/avg. of 32 hits total at present which to me is still too many. The backtesting work here has helped tweak the scans to fewer hits but going onto that first site you mentioned makes my user ID "handle" seem out of touch to those folks scope of work and maybe I should change it to "greenhorn" in comparison. But I have always logged my hits by scan, dated, added vol. increase, HOD, P/E, sector notes as I've went along and am using the data to put scans into an "excel" chart so as to match performance to the DOW (my reference median line) and establish if there's a trend where by certain scans work better during certain market conditions and use this as a concentration point till I get the scripts tweaked even better. The (8) scans have been logged religiously daily since only 6/6/03 and just for fun I recorded the price increases of scan #1's hits over this past (2) week period and it showed out of (89) hits in (10) trading days...(4) ended up being losing stocks so far when verified using YHOO historical check. Of the earlier hits, (26) peaked in one week and (27) peaked in two weeks, so the scripts seem valid but the scope is still too large which leads me back to charting these to narrow my focus. I'll write again when I have a broader range of data and some historical figures. Thanks again for posted sites.
|
defghca
150 posts
msg #27016
- Ignore defghca |
6/23/2003 7:06:34 AM
Stockfetcher team
after reading the latest posts in the filter section i think you will agree we need to device a better performance reporting system.
|