holygrail
72 posts
msg #27001
- Ignore holygrail |
6/22/2003 6:57:16 PM
Show stocks where Close gained more than 20 percent over the last 4 weeks and Average Volume(5) is above 50000 and the rsi(2) is below 70 and thr rsi(2) has crossed above rsi(3) and close is between 0.01 and 1.0 and the historic volitilty(5) is above 70 · remove · edit
|
no4j
45 posts
msg #27026
- Ignore no4j |
6/23/2003 6:24:13 PM
Love the backtesting. Lets say I wanted to buy just one stock on each day that the filter returned at least one. What decision making process would you put into place for choosing the one you would want to buy?
Dave
|
dwh0000
5 posts
msg #27028
- Ignore dwh0000 |
6/23/2003 8:30:07 PM
I backtested this for February of this year and this filter didn't look that good. Maybe it only works in bull markets.
|
paulsam
2 posts
msg #27031
- Ignore paulsam |
6/23/2003 9:47:43 PM
how do you backtest on stockfetcher?
|
xplorer
257 posts
msg #27036
- Ignore xplorer |
6/23/2003 10:49:19 PM
The way I back test is to edit the given filter and add to the end "and offset 1 day" .... then change it to "and offset 2 days" ... or whatever time frame you are looking for your filter (or "fetch") to work within.
Good luck !
|
defghca
150 posts
msg #27040
- Ignore defghca |
6/24/2003 6:47:18 AM
of course it will be the best 1 month filter , you set it as a precondition of at least 20% growth "Show stocks where Close gained more than 20 percent over the last 4 weeks " get real !!
|
risjr
229 posts
msg #27041
- Ignore risjr |
6/24/2003 10:22:56 AM
This filter back tests dynamite in a bull market. It's great! Once more, it is excellent beyond 30 days as well. Many with triple digits in 60 days or less!And it has a large percentage of big gainers, not just a few!
Yo Holy Grail
What is your best filter in a bear market? Looks like this coud be a vicious one
Thanx
Ron
|
karennma
8,057 posts
msg #27071
- Ignore karennma |
6/25/2003 8:50:54 PM
HolyGrail:
Kudos to you! That filter is awesome. For my purposes, I removed the rsi(1) x'g over the rsi(2), and it works quite well.
THANKS.
Any ideas for shorts?
BTW - I don't know what value the "historic volatility" is, but it WORKS!!
THANKS!
|
dmindy
42 posts
msg #27076
- Ignore dmindy |
6/25/2003 9:43:11 PM
ok everyone... step back a minute here.
I'm not commenting on the filter.. maybe it is the best ever, I never ran it but PLEASE keep in mind...
"date offset 30" does NOT give you a good backtest for "periods of thirty days". All it does is give you what happened this last 30 day period.
In order to get that you'd need to test for thirty day periods starting at various dates in the past. Stockfetcher can't do that (as of yet).
If you bump it up and do a date offset 60 or 280 (or whatever) the percentages reflect what happened from that date until NOW. Also not a good back test.
The only way to back test for thirty days via stockfetcher is to do a date offset 30 every day for a year or two AND THEN report the results. Not very effective.
Date offset gives you a GREAT "rough starting point" for possible winning filters but shouldn't be your only source of "back testing"
There are plenty of good filters, and given enough time (20 days, 30 days, a thousand whatever) most of them will look awesome.
Practice Due diligence before labeling something a 100% winner.
|
xplorer
257 posts
msg #27077
- Ignore xplorer |
6/25/2003 10:35:59 PM
dimindy...
If you are a trader... and enjoy the excitement and profitability that short term trades offer then using "offset X days" would be fine (x being the number of days you want to hold it) ... and yes you should test it for a range of days within your trading timeframe. Though I do agree this is limited testing as your window of testing is during the current market conditions ...and we all know that can change in a heart-beat ....but it is the best way I can see to test ....right now.
These guys that created this site are awesome ...and always ready to listen too ...and respond to suggestions ... but what would be a better way to back test ??? Maybe "Show results from 1 month ago and offset "x" days" ??? Or does anyone have a better possibility ?
|