StockFetcher Forums · General Discussion · OBV crosses.... what is the correct syntax?<< 1 2 3 >>Post Follow-up
jclaffee
81 posts
msg #39210
Ignore jclaffee
11/30/2005 2:08:47 AM

ImperialWhazoo:

Thanks for kind remarks. Please don't interpret my slow response as evidence of lack of interest in your work. What it IS evidence of is that I have my hands full getting ready for each trading day. I use weekends to pursue "interesting things" that I haven't had time to sort out during the work week. BTW, I use the work done by Yepher on the SF commands. . .I think it a superior instruction book. . .you can Google it up with "stockfetcher commands Yepher".

Jim


jclaffee
81 posts
msg #39212
Ignore jclaffee
11/30/2005 4:18:34 AM

ImperialWhazoo:

I don't have any answers for you but I'll show you the small amount of foolin' around I did tonight. . .it might set off something. . .

Fetcher[Average Volume(40) is above 300000
and Close is between 0.05 and 5
and MA(40) has been increasing over the last 3 days
and Slow Stochastic(12,3) Fast %K is below 50
and set{recent, OBV(12)}
and set{delv, OBV - recent}
and delv is above 0
and add column OBV
and add column recent
and add column delv
and do not draw delv
and draw recent
and draw OBV on plot recent
and draw MA(7)
and draw MA(12)
and draw MA(40)
and draw Bollinger Bands (20,2.0)
]



I just wanted to get the various elements up where they could be seen on the results and chose column headings that had relatively few characters.

As I said, there's no gold here -- there's nothing but rocks -- but perhaps an idea will peep out!

Jim


jclaffee
81 posts
msg #39213
Ignore jclaffee
11/30/2005 8:49:49 AM

A few more "rocks" for ImperialWhazoo. . .

1. I created a watchlist which I named "IWsF" (ImperialWhazoo'sFilter).
2. I ran the filter in my previous message, backed it up 15 days (to get it before the November 11 date you had previously found interesting), and loaded the watchlist with the results.
3. I set up the following filter to follow those stocks in watchlist IWsF from day-to-day:

Fetcher[set{recent,OBV(12)}
and set{delv, OBV - recent}
and delv is above 0
and add column OBV
and add column recent
and add column delv
and do not draw delv
and draw recent
and draw OBV on plot recent
and draw MA(7)
and draw MA(12)
and draw MA(40)
and draw Bollinger Bands (20,2.0)
and apply to watchlist(IWsF)
]



4. I examined the issues in the watchlist over the span 25 days ago thru yesterday, 11-29. When I claim to have "examined", I mean in the most cursory fashion!

5. I still think there's something funky about "OBV(12)": there seems to be a fixed relationship between OBV and OBV(12) for each issue. They each change
but the distance between them does not. In the terms of my tweak of the filter,
"delv" is a constant.

6. I wonder if there's anything to be extracted from x-overs of MAs of Volume? For instance, MA(100) x MA(10), or similar.

FWIW

Jim



ImperialWhazoo
26 posts
msg #39220
Ignore ImperialWhazoo
11/30/2005 11:01:47 AM

jclaffee -
Thanx

1. I use the yepher manual too 'cause it is better
2. I agree there is something funky about OBV(12).

Frankly, I couuld use the default, which is 10 (if my recollection hasn't fail me). in any event, my messing with this led me to conclude that there is indeed something funky about OBV(12).... and OBV(10)....and OBV(7).... and so on.

One guy I listen to on the internet radio uses 10 periods as his choice. Another uses 22. A third prefers 32. I prefer to use 3, 7, 12, & 40. In this case OBV(12) works better than 3, 7 or 40. My point is that you can take it or leave it as regards 12 but the funky thing is still the stumbling block as regards screening for OBV upward crosses.

By the way, I'm trading right now so I'll look at your screens later tonight.
Also, like you, early on in my pursuit of figuring out if the DoubleUp-Cross could be used in a meaningful filter, I also put the various column headings and graphically presented lines on the charts in order to better see the lay-of-the-land.

Heres what I've deduced thusfar.... if I could figure out how to properly structuire the syntax, I could better pin this down:
a.) when the Stochastic cross I'm looking 4 happens below 35, this in meaningful. Above 50, I have less certainty. Thats one of the backtests I'd perform.
b.) The cross of the red & blue OBV lines can actually be either a cross or the narrowing of the distance that the blue and the red lines. This is another thing I'd like to bAcktest... whether (when the blue OBV line is already above the red OBV line) a touch or near touch is equivelent to a cross from below.
c.) the trading plan I use today that seems to work but which I'd like to refine is to have a working list of tickers that are approaching firing off a signal. What I've been able to surmise so far is that either upward cross can happen first. If a cross happens on OBV, for example, and the Stochastic is declining but starting to curl up (not yet having crossed), then this ticker goes on the watch list. The opposite also applies where the Stochastic has already crossed and the OBV appears to be about to happen. And the obvious one where both crosses happen the same day also works (duh!), in which case you don't put the ticker on the watch list... you evaluate it as to whether to buy it the next morning.
d.) the two crosses can be close to each other or up to 5 days apart, or so it appears to me thusfar. But again, without the ability to properly structure a syntactically correct filter for backtesting, I'm working manually at present and have not been able to be systematic. The key when several days separates them is that the one that fired first can not "decline" in its "quality" as a cross. It, for example, can not cross back downwards or the efficacy of the original signal evaporates.
e.) At present, I buy at or near open on the next day after there have been both crosses confirming. I hold until the basic candlesticks indicate that the short term upward trend has begun to erode.

Thats where I am at present. Oh, for another example of what I'm talking about, look at VYST based on my settings and the crosses that happened Nov 2.... Voila! A DoubleUp cross event ALSO happened on about Oct 17 on VYST and the thing that I have deduced that is the rule is that the white candle on Oct 17 was FAR TOO TALL!! to work... in fact, my rule has become that a major white candle causes a "distortion" that forces these dual upward crosses to happen and that the very size of the candle in relation to the normal size of the candle should be a warning to hold off on trading the type of DoubleUp-Cross event that happened on Oct 17... and notice that the Nov 2 DoubleUp Cross (which was hugely profitable where the Oct 17 trade would not have been at all) did not involve any abnormally large white candles. Again, backtesting is needed to pin this down.

I'll take a look at your kind comments after the market today

Keener.... the ImperialWhazoo


ImperialWhazoo
26 posts
msg #39232
Ignore ImperialWhazoo
11/30/2005 7:46:23 PM

Hey dangreene -
Take a gander at this!!!
I decided to try your first suggestion again on the thought that, despite the fact that your CEMA and the OBV(12) lines only APPROXIMATE each other. And I kind of lectured myself on how silly I was for insisting that APPOXIMATE is not good enough. I reasoned that even though the CEMA was slightly off, it was usable.
Well.... take a gander at this! It turns out that the difference is sometimes HUGE. In fact, if you replicate what I'll describe following, you will see that, for ticker ABTG, a cross using the CEMA happens on Nov 5 but the cross on the OBV EMA(12) happens Nov 18.... a full 8 trading days different!!! Not exactly a SMALL difference. I'd have to conclude that the CEMA(OBV(12),12) and the "native" OBV(12) graphs are sometimes not even approximations of each other.

So, check this filter out (for end of day 11/30/05, by the way)

Fetcher
[set {obv12, OBV(12)}
set {obv12ema, CEMA(OBV(12),12)}
Show stocks where Average Volume(40) is above 300000
and price is between .05 and 5.01
and obv12 crossed above obv12ema in the last 5 days
and Slow Stochastic(12) Fast %K crossed above Slow Stochastic(12) Slow %D
and draw OBV(12)]

Now, please don't think I'm critical of your contribution in any way because I I actually decided, as I said at the top, that I was being silly for making so much out of the differences in the two lines. And also, as I think I mentioned in an earlier blog to you, I also tried the CEMA many moons ago and, until I got back into it because of your contribution, I had forgotten why I'd discarded using it. Now, looking at ABTG using the above filter, you too can see what makes me so grumpy about this.

Thanx etcetera...
ImperialWhazoo


ImperialWhazoo
26 posts
msg #39233
Ignore ImperialWhazoo
11/30/2005 7:58:53 PM

OOPS!!
I just noticed I said Nov 5 and Nov 18 and called it 8 trading days. Well, NOT!!

Use this filter and look at ABTG:

Fetcher [set {obv12, OBV(12)}
set {obv12ema, CEMA(OBV(12),12)}
Show stocks where Average Volume(40) is above 300000
and price is between .05 and 5.01
and obv12 crossed above obv12ema in the last 5 days
and Slow Stochastic(12) Fast %K crossed above Slow Stochastic(12) Slow %D
and draw OBV(12)]

The actual crosses happen Nov 11 ( a down cross in this example) and Nov 17 (also a down cross). Still 6 full trading days apart... a HUGE difference! Obviously not usable approximations of each other, as I had started out assuming. Thus, CEMA(OBV(12),12) and OBV(12) are not substitutes for each other which means that I'm still stuck trying to figure out how to screen for the blue line crossing the red line when using OBV(12) or OBV(7) or whatever.

Sorry for the error!!
Regards etcetera...
ImperialWhazoo



dangreene
133 posts
msg #39235
Ignore dangreene
11/30/2005 10:19:50 PM

ImperialWhazoo,

Thanks for your continuing struggle to reach a solution to your approach. I still believe that what you are attempting to achieve has real merit and am interested in pursuing it further.

However, I agree with jclaffee when he said, ---- "I still think there's something funky about "OBV(12)".

Since there seems to be little interest in this forum from others (except for jclaffee) I would prefer to continue this privately.

Please email me at dangreene@yahoo.com as I have come up with something that should be of interest to you.

I may not be able to respond until late tomorrow, 12/01/2005 as I have other commitments so bear with me.

jclaffee, because of your participation and work to help in this effort, I invite you to email me also since you seem to be interested in this approach.

email: dangreene@yahoo.com



jclaffee
81 posts
msg #39237
Ignore jclaffee
12/1/2005 3:03:24 AM

Hi, dangreene. . .

Thanks for the invitation to join you and ImperialWhazoo in an e-mail exchange about this thread. I'm going to decline your offer because 1) I don't think I have much to offer and 2) I lack a real compelling interest in the issue.

I am a swing trader and I don't see opportunity for me in the charts we've looked at nor do I have a high opinion of my SF language abilities.

I DO continue to think that a fundamental problem is the effort to put a time period on a moving sum whatever that period may be. I would look for a proxy indicator that would do the trick: how about Money Flow Index which combines input from price and volume? Reference yepher p. 173.

Happy filtering to you both!

Jim




mehrdad25@sbcglobal.net
11 posts
msg #39239
Ignore mehrdad25@sbcglobal.net
12/1/2005 7:25:07 AM

I had this exchange about a year ago on this board:
http://stockcharts.com/education/TradingStrategies/lastStochastic.html

K39 - The Last Stochastic Technique
A study published in "The Encyclopedia of Technical Market Indicators" found that some very good signals were given by an unsmoothed 39 period stochastic oscillator (K = 39, no signal line). A buy signal is generated when K crosses above 50% and the closing price is above the previous week's high close. Sell and/or sell short signals are created when the K line crosses below 50% and the closing price is below the previous week's low close.

Anyone knows how to write the correct factors in the sreen box (Stockfetcher.com) to screen for the "The Last Stochastic Technique" described above?
1- Unsmoothed 39 period stochastic oscillator (K = 39, no signal line)
2- On Balance Volume (OBV) indicator along with a 30 day MA added as a signal line.
I hope I am making some sense here!
Thanks in advance.





robdavis 8/20/2004 4:59:19 AM

Mehrdad,

Fetcher[/*TO BUY*/
Show stocks where
close is between 1 and 250
and average volume(90) is above 50000
and volume 30 day low is greater than 50000
and Slow Stochastic(39,1) Fast %K has crossed above 50 in last 1 day
and close is above close 1 day ago
and close is above close 2 days ago
and close is above close 3 days ago
and close is above close 4 days ago
and close is above close 5 days ago
and draw Slow Stochastic(39,1)
and draw Average OBV(30)
]




Fetcher[/*TO SELL OR SHORT SELL*/
Show stocks where
close is between 1 and 250
and average volume(90) is above 50000
and volume 30 day low is greater than 50000
and Slow Stochastic(39,1) Fast %K has crossed below 50 in last 1 day
and close is below close 1 day ago
and close is below close 2 days ago
and close is below close 3 days ago
and close is below close 4 days ago
and close is below close 5 days ago
and draw Slow Stochastic(39,1)
and draw Average OBV(30)
]



I hope this helps,

Rob
----------------------------------
In most cases, I get a cross over signal AND I don't know why I am getting it BUT that's what I am looking for too. Hope his helps.
regards
Mehrdad


BFreshour
139 posts
msg #39241
Ignore BFreshour
12/1/2005 8:50:51 AM

Thanks for sharing those filters. I've been going over them and they look to be able to return some winners.


StockFetcher Forums · General Discussion · OBV crosses.... what is the correct syntax?<< 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.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus