StockFetcher Forums · Filter Exchange · Bollinger Squeeze << 1 2 3 >>Post Follow-up
yepher
359 posts
msg #31689
Ignore yepher
4/11/2004 7:05:33 PM

Sorry if this has already been posted.

I have been reading Bollinger on Bollinger Bands and thought I would share the filter that is based on the formula he provides. This will give you the stocks whose bollinger bands have reached a new low compression level of the past 6 months. (Side note: I generally don;t like using the set command because to me it makes the filter harder to read but I think in this case it simplifys the filter)

Fetcher[set{top,upper bollinger(20)} set{bottom, lower bollinger(20)} set{squeeze, top - bottom / ma(20)} AND squeeze reached a new 6 month low]

If anyone has any suggestions on improving it I would love to hear about it.

-- Yepher

yepher
359 posts
msg #31690
Ignore yepher
4/11/2004 7:19:48 PM

Here is anoter variation of th same filter. Sorts by the most compressed

Filter[
set{top,upper bollinger(20)}
set{bottom, lower bollinger(20)}
set{squeeze, top - bottom / ma(20)}

AND squeeze reached a new 6 month low
AND close near upper bollinger(20)
AND price between 1 and 10

AND sort by column 5 ascending
AND chart-time 68
]

yepher
359 posts
msg #31691
Ignore yepher
4/11/2004 7:24:12 PM

Repost of same filter to see if I can get it clickable and still have multiple lines.

Here is anoter variation of th same filter. Sorts by the most compressed

Filter[_
set{top,upper bollinger(20)}_
set{bottom, lower bollinger(20)}_
set{squeeze, top - bottom / ma(20)}_
AND squeeze reached a new 6 month low_
AND close near upper bollinger(20)_
AND price between 1 and 10_
_
AND sort by column 5 ascending_
AND chart-time 68]

 dangreene229 postsmsg #31697- Ignore dangreene 4/12/2004 7:27:07 PMyepher, I would suggest scanning for a specific value for the "squeeze" rather than looking for a new 6 month low. VIZ: set squeeze, top - bottom / ma(20)} AND squeeze is below .20 My thinking is that there could be a couple of squeezes in a 6 month period and todays value might be .14 while the previous values might be .13 or less. With a value this low, who is to say that the most current squeeze might not turn out to be the most profitable? Thanks for sharing your scan as I think it has a lot of potential. dangreene
cegis
235 posts
msg #31718
Ignore cegis
4/13/2004 8:23:37 PM

yepher,

First, thanks from me as well for the filter idea...

Second, to make the filter a clickable link, use "Fetcher", not "Filter", as in:

Fetcher[
set{top,upper bollinger(20)}
set{bottom, lower bollinger(20)}
set{squeeze, top - bottom / ma(20)}

AND squeeze reached a new 6 month low
AND close near upper bollinger(20)
AND price between 1 and 10

AND sort by column 5 ascending
AND chart-time 68
]

Thrid, dangreene may have a point. I haven't played with this filter (yet), so I can't say for sure. Seems to make sense, though. (And not so much as the recent squeeze being MORE profitable, as just being profitable AGAIN!)

Lastly, as an idea that just popped into my head, maybe it would be interesting to do another filter where you use something like "set{days010,days(squeeze is less than 0.1,200)}" and "set{days015,days(squeeze is between 0.10 and 0.15,200)}", etc. then "set{sumdays,days010 + days015}"... and finally "sumdays is greater than zero". This will give stocks that have had at least one, but possibly several, squeezes over the 200 day time frame, and may be useful in determining what a good squeeze value might be...

HTH, and thanks...

C

yepher
359 posts
msg #31726
Ignore yepher
4/13/2004 11:30:46 PM

First of all I should say, I am a big fan of the people who contribute to this forum and have been watching for some time now. I am glad to finally have something that I can contribute.

dangreene,

- I guess I could use the syntax "near" for the 6 month lows if the filter seems to be too tight or missing good canidates. Yes?

Cegis,
- Ahh yes Fetcher not Filter. Sometimes I just overlook the obvious DOH!!!

- I like the final idea you proposed. I think a varation to that which I will research is: if currently in a squeeze now and the previous squeeze of same relative magnitude performed well in average price(N) then does that provide any addition statistical confidence and predictability. This is purly looking for magnitude of breakout and not direction.

- This filter will not do well if you use the performance analysis because it does not attempt to define direction for movement. It only tries to find out that there is going to be a movement.

My thinking on the filter,

My first pass was to impliment as exactly as possible to filter described in Bollingers book and here: http://www.investopedia.com/articles/technical/04/030304.asp

I will continue my research on Bollinger squeeze. From my study of this filter it seems you get a nice price shift within the next 2 - 7 weeks. I don't think Bollinger band really indicates direction so I am working on a complimentry divergence filter to indicate direction.

From this bollinger filter I want to derive the following:
A. Bollinger squeeze between some range
B. Bollinger band just started to loose some compression
C. Divergence of oscillator

I have parts A & B And I have been studing all the posts on divergence as well as the SF manual but I have not quite figured out how to accomplish part C. I am thinking I need to use linear regression for the lows (or highs depending on what Long or Short prospect) of the oscillator and tie that to the linear regression of price somehow. I am not convinced that the divergence discussed in the forum is the same as the one I am looking for. Need more research on my part :)

The oscilator that I hope to use is one with a lot of smothing and not suseptible to hiccups when data falls large price deviation fall off the end of the window. More smothing that Stoch would be great.

What I hope to find is a stock that has a good chance of movement and with some level of direction prediction. It would be nice to also get some insight into head fake and how to detect it.

-- Yepher

dangreene
229 posts
msg #31733
Ignore dangreene
4/14/2004 10:51:21 AM

yepher,

After looking at your scan more closely I would also like to suggest that you
consider eliminating "AND close near upper bollinger(20)_".

My reasoning is that first of all it limits the number hits you get back and
secondly, when the bands have contracted to this extent just how relevant is
this requirement?

You said,
"The oscilator that I hope to use is one with a lot of smothing and not suseptible to hiccups"

For what it's worth, I use slow stochastic(39,3) and wait for fast %K to cross above 50 as an indicator of strength and likely direction. Once on a watch list I then like to see a positive DMI crossover.

At first blush it seems that this scan will do well when the market is trending up. Check a chart for the nasdaq-100 and run your scan with a days
offset to give you results for Tuesday March 23, 2004 as an example.

Cegis, glad to see your input here. It takes me a while to digest some of your code but I end up seeing the value in it.

Hope we can all continue to contribute and come up with a productive scan.

dan greene

dangreene
229 posts
msg #31739
Ignore dangreene
4/14/2004 4:24:11 PM

yepher & cegis,

Here is an extreme example of the slow stochastic(39,3).

Notice how it remained at a high level on the following stock.

http://stockcharts.com/def/servlet/SC.web?c=HIET,uu[h,a]daclyiay[db][pi!d20,2!f][vc60][iUh39,3!Ul14]&pref=G

HIET showed up on my version of the BOL BAND Squeeze scan for Thursday April 1, 2004.

See how HIET traded in a narrow channel from Feb 2 until Mar 4 & Mar 5, at which time Stoch(39,3) bumped up a notch along with a bump up in a 14 day +DI & ADX.

Then after hitting a high on Mar 15 it traded in a channel again until the BB squeeze on Apr 2. Once again the Stoch(39,3) bumped up along with the +DI(14) on Mar 6.

Here is my version of the scan:

Fetcher[
Show stocks where market is nasdaq
and set{top,upper bollinger(20)}
set{bottom, lower bollinger(20)}
set{squeeze, top - bottom / ma(20)}
AND squeeze is between .05 and .20
and average volume(30) is above 150000
and Slow Stochastic(39,3) Fast %K is above 50
and close is above 1
and sort column 5 ascending
and days offset is 0
]

If you guys feel it has merit maybe you can find a better way to incorporate the Stoch(39,3) into the BBSqueeze scan, or maybe even develop a new scan to find channeling stocks where the Stoch(39,3) remains at a high level.

It also seems that RSI(2) could be an important indicator if it reads ZERO a day or two before the squeeze date or on the squeeze date. I am going to continue to check on this.

dan greene

yepher
359 posts
msg #31742
Ignore yepher
4/14/2004 6:05:56 PM

Dan,

Thank you, for all the input this is good stuff. I like the direction you are going. I am going to spend time with this filter you have and see how far it moves toward where I think this can can go. To me a true test of the filter would be that it returns very few results when the market is stagnant and many results when the market is on the move.

Originally I was not going to worry about ADX since I thought the fact that there was bollinger compression ADX would already be showing oscilator. I need more thought on this point.

I realized last night that I did not mean to say "Linear Regression" of lows on oscillator but the slope of the lows on the oscillator to help judge direction. Sorry if that caused any confusion.

-- Yepher

dangreene
229 posts
msg #31744
Ignore dangreene
4/14/2004 6:34:40 PM

Yepher,

You said,

"I realized last night that I did not mean to say "Linear Regression" of lows on oscillator but the slope of the lows on the oscillator to help judge direction. Sorry if that caused any confusion."

I tried adding "slope of slow stochastic(39,3) above 0" to the scan but too often the stoch had already turned down yet satisfied the condition and I find it unproductive to enter a trade when the stoch(39,3) is turning down.

dan greene

 StockFetcher Forums · Filter Exchange · Bollinger Squeeze << 1 2 3 >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.