macer
22 posts
msg #53771
- Ignore macer |
8/2/2007 12:52:00 PM
What is the most accurate filter that you can produce to predict the next day price movement?
To keep it fair/realistic a few ground rules are:
* Must produce a decent amount of trades - by that I mean at least 3 a week
* Price must be more than $1 - No penny stocks
* The entry price is open and exit price is close
* Trades have one day holding period
* Filters may have a short or long approach
* Results should be ordered highest to lowest probability - Not essential
Let’s see who can produce the highest win/loss ratio
Macer
|
luc1grunt
622 posts
msg #53779
- Ignore luc1grunt |
8/2/2007 4:27:36 PM
You start :)
|
macer
22 posts
msg #53781
- Ignore macer |
8/2/2007 5:54:42 PM
Yeah, should have seen this coming and prepared something but 65% W/L is about all I've got. Here's the filter for your amusement:
/* Short williams rsi */
lrs(60) < 0
williams %r(3) < williams %r(3) 1 day ago
rsi(2) > 99
rsi(2) > rsi(2) 1 day ago
slope of rsi(2) has been decreasing for 2 days
close < open
price > 5
average volume(90) > 1000000
I have "borrowed" some ideas form previous posts, hope no one minds.
|
nikoschopen
2,824 posts
msg #53787
- Ignore nikoschopen |
8/2/2007 9:15:34 PM
What is the most accurate filter that you can produce to predict the next day price movement?
It's unthinkable that you even asked such a question. C'mon now, would YOU share such a filter if you had one?
:)
|
jrbikes
624 posts
msg #53788
- Ignore jrbikes
modified |
8/2/2007 9:23:26 PM
No Penny stocks? Im out!
Does a 29% win rate with a ROI of 540% count? LOL
take your worst filter and do the exact opposite!
|
macer
22 posts
msg #53795
- Ignore macer |
8/3/2007 6:31:56 AM
“It's unthinkable that you even asked such a question. C'mon now, would YOU share such a filter if you had one?”
Nikoschopen, I don’t expect anyone to bang a filter that is 100% accurate on the table.
I was wondering what the best W/L ratios that have been achieved by others. My reasoning behind my post was to see what accurate filters had in common and learn from them. If the goal is to find accuracy of selection and not focus on ROI compounding will take care of the rest (in my opinion).
Jrbikes that probably good advice. No need to work hard on developing new filters just let the inevitability of failure take hold then run with it.
Macer
|
conyeuchua
83 posts
msg #53800
- Ignore conyeuchua |
8/3/2007 11:07:55 AM
Search the forums for Alf44's "work of art" filter posted on July 1, 2007. It is very extensive but Alf44 explained that filter very well. There are some components in his thinking and coding that may help you answer your own quest.
I don't have the link to it handy. Good luck.
|
jrbikes
624 posts
msg #53828
- Ignore jrbikes
modified |
8/3/2007 10:41:09 PM
LOL, thats sage, Man! (Macer)
I hate failure, I was just thinking about something someone else said, in developing a mechanical trading system.
Man, that was good! LOL
|
rwgamer
17 posts
msg #53848
- Ignore rwgamer |
8/5/2007 11:59:51 AM
It seems like most people on this forum focus too much on Win ratio. Check out Van Tharp's book "Trading Your Way to Financial Freedom". He explains expectancy, and that you can make more money with a 40% win ratio than 80%. It's a 500 page book, so I can't wrap up the concepts in a nice package...
This filter produces a nice Reward to Risk ratio. The win ratio is usually around 30%, but if you were actually trading this you wouldn't trade everything that came up in the filter, you would look at support and resistance, and other things that the filter can't detect. In real trading, your win ratio would probably be closer to 50%.
This filter looks for volatile stocks that have moved lately.
/* Big Move + Volatile */
show stocks where the average day range(10) is above 15 percent and close price is between 1 and 40 and volume is above 300000
/* Long */
and RSI is between 10 and 45
/* Short */
/* and RSI is between 55 and 90 */
|