wedad73
2 posts
msg #34019
- Ignore wedad73 |
11/20/2004 8:41:41 AM
Hello freinds ,
does any expert one tell us a filter which find stocks which make a positive divergence with its macd (26,12) while the stock in downtrend when its make its positive divergence ?
these kind of stocks as you know make a huge movement if one could find them and play buy game with them ..?
thanks alot
|
xplorer
257 posts
msg #34029
- Ignore xplorer |
11/20/2004 9:24:44 PM
I pulled this from SF ... but I am not able to give credit to the creator...
... hope this helps !
|
yepher
359 posts
msg #34040
- Ignore yepher |
11/21/2004 2:44:56 PM
wedad73,
xplorer's post may answer your question perfectly. I thought this would be a really good time to bring up divergence the way I find it most useful.
I have tried several time to produce a general purpose way to find divergence of a given indicator. I have yet to personally solve the problem. Here is what I mean by divergence with the hope that by describing it I inadvertently figure it out or someone else has the answer that they would share.
I have seen many posts about divergence through the years on this forum but have yet to see any that find the divergence I mean.
Here is a picture of the divergence I mean:
http://www.stockcharts.com/education/IndicatorAnalysis/Images/rsi-dellss.png
or maybe this is a better image:
http://www.chartfilter.com/reports/c39_files/c39-1.gif
(it is only a coincidence that these two happen to use RSI measures)
When I talk about divergence I mean a particular price event and indicator event. I want to find the previous time that occurred and measure the slope between them. I want to know the slope of the indicator points and price for the two points relate those two values together.
As far as I know this is only available as manual line studies of a chart.
I believe the problem is partially solved with the "slope" syntax. It would be especially useful if slope could take a variable that was formed from "days since" or "count" commands.
|
james549
53 posts
msg #34041
- Ignore james549 |
11/21/2004 3:36:36 PM
yepher
Is this what you are looking for?
|
yepher
359 posts
msg #34042
- Ignore yepher |
11/21/2004 6:53:29 PM
james549
Thanks for the effort, but that is not quite what I am looking for. The divergance I want would need a variable for the input to the slope command.
Here is psuedo code for what I would like:
var point1 = days since peak in indicator
var point2 = days since previous peak in indicator
var divergance = slope of indicator betwen point1 days ago and point2 days ago
To keep the example simple I did left that part out that would tie this information back to the price chart.
|
wedad73
2 posts
msg #34048
- Ignore wedad73 |
11/22/2004 8:15:59 AM
xplore and all freinds,
thanks for your help .. but stil I have a question , should I put the offset 0 for the daily search or should I change it ??
thanks alot ..
|
xplorer
257 posts
msg #34065
- Ignore xplorer |
11/22/2004 7:18:10 PM
the "and date offset is 0" is for backtesting...
.... it doesn't add to the filter. ... though I keep it set to zero, then when the market sentiment changes it is an easy "re-test". The problem with using this, is that is always tests up to the present day... so if you have a filter designed for a 5 day or less hold, it won't provide much useful info, test further back.
Also, be careful, it is easy to mis-interpret the results, as when the markets up, you filter can show alot of 20 - 30% returns in a day... the mother lode... ...and any futher testing will also be biased to the current market conditions. Take a look my posts within this "Filter Exchange", "BACK TESTING - do it right !". I believe a have a reliable way to back test ... not real convenient ... but it does seem to work.
... so back to your question, by changing the offset to different number of days you can see how the filter performs (biased to todays market sentiment)...
|
xplorer
257 posts
msg #34066
- Ignore xplorer |
11/22/2004 7:41:50 PM
Yepher
My take on "divergence" is usually two parameters that have opposite slope (negative and positive),,, and price being primary. Given that, slope of each of the parameters should be relatively eacy to compare. Although I am not fully understanding how SF calculates slope ...
|
yepher
359 posts
msg #34067
- Ignore yepher |
11/22/2004 11:13:10 PM
xplorer,
I know how to get the slope of the points. The part I am not clear on is how to find the points in a generic way. For instance a phrase such as "10 day slope of some indicator" will not work.
What I would like to do is have some filter that would find the slope between to give offsets for some indicator. See links from previous post for example of the divergence I would like to find.
|
luc1grunt
622 posts
msg #61453
- Ignore luc1grunt |
4/14/2008 9:12:55 AM
"pop" great topic
|