SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Strategies & Market Trends : Technical Analysis- Indicators & Systems

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Bill Sandusky who wrote (2044)7/15/1997 11:55:00 PM
From: Paul Beattie   of 3325
 
Explorations and Exploits with BNS Indicators in MetaStock.
This is a "Work in Progress". You and I have discussed this material, but I want to post it so others can follow along. It is based on the status of investigations as of July 11. I don't consider it complete or perfect.

This is a summary of where I'd progressed up to TAINJ. My thanks and acknowledgements to you, Bill, for starting all this.

I'm going out of town and won't be able to respond or answer questions until after July 26. In order not to experience TA/SI-withdrawal tremors during that period, I'm taking the TAINJ material with me!

Bill, Thanks for making TAINJ a special event for all of us.

- Paul

Background
In addition to CCI(13) and MACD(5,35,4), BNS makes use of the interrelated movements in the following 4 indicators:
ú Avg_Price
ú Chaikin (Inverted)
ú RedLine
ú StdDev

Their movements are easiest to see when they are overlayed in a single window. Bill uses TradeStation, which allows him to scale these indicators to the database for comparitive purposes.

I use Metastock for Windows, which allows for overlaying of indicators, but will repeatedly rescale them based on data in the window displayed. This results in indicators which slide up and down in relation to one another whenever the date range in the window changes.

P-Indicators
I tried to find an approach which let me see the indicators clearly separate from one another when they were at extremes. My approach was to map each of the 4 indicators to "oscillators" which range from 0 to 1. The value of each oscillators (P-Blue, P-Yellow, P-Red, P-White) is the deviation of the indicator from it's 28-day extreme, expressed as a percentile. A 28-day high maps to 1, a 28-day low maps to 0. The oscillators don't slide up and down relative to one another as the screen is scrolled/zoomed. However some information is "lost" when an indicator makes a new 28-day high or 28-day low. (Also, if an indicator stays horizontal for a 28-day period, the oscillator will ultimately float up to 1).

Using 4 P-indicators which all share the same scale, it is not "possible" to scan for relationships (A>B, A<B) and crossovers among them.

Mapping of Indicators:
Chakin (inverted)...............P-Blue
AvgPrice........................P-Yellow
RedLine(-Proxy).................P-Red
StdDev..........................P-White

Summary of Findings/Interpretations about P-Indicators:
They show clearly the relative behaviour of BNS indicators around extremes. They point out several patterns (some of which are difficult to test using regular system testing methods), including: sudden moves down of P-Blue, especially when combined with sudden moves up of P-Yellow.

Visually, which indicator is leading?
Chaikin (inverted)......generally moves a day or two before AvgPrice.

AvgPrice................can be a leading indicator to price moves

RedLine.................Proxy I use [Stoch(8,5)\ is not as sensitive as RedLine

StdDev..................Doesn't really lead anything. It follows trends up and down, but doesn't care which. Tough to find consistent visual behaviour relative to other indicators. Visually, it sometimes looks like it is "setting a support for a climbing Price". (when it slowly climbs from an extreme low).

BNS P-Yellow System
This is not the BNS system. It is simply based on my "derivative" of AvgPrice. Of my simple efforts to system test with AMTX, the best performing system that I have found is fairly simple:

Buy Long - when P-Yellow is above .7
Exit Long - when P-Yellow is below .05

using the following formulas:
AvgPrice: (((H+L+C)/3)-Mov(C,35,S))/.015*Std((H+L+C)/3,35)
Yes, this is missing the brackets in the denominator.

P-Yellow:
1-Div(HHV(Fml("AvgPrice"),28)-Fml("AvgPrice"), .001+HHV(Fml("AvgPrice"),28)-LLV(Fml("AvgPrice"),28))

I have obtained interesting results when testing on AMTX and other stocks. It is "trigger-happy" on the RHS of the mountain or in a basing or slowly decling period. I would not use it's signals in these situations. Otherwise, it makes some good entry and exit timings.

Other comments (for what they are worth):
System testing BNS will not be easy, even using the P-indicators.
It is not based on levels and crosses only. At good buy opportunities, the indicators behave consistently in relation to one another in a way that I notice visually. It's tougher to test for this mechanically.

When P-Yellow turns up sharply before P-Red, it often signals a strong move up, and Redline follows it within a day or so. When P-Yellow turns down sharply before P-Red (percentile of Redline), it is often the end of a trend. However, if P-Yellow stays high and P-Red turns down, slowly followed by P-Yellow, I wondered if it's a sign that there is more steam in the uptrend and it may be just taking a breather.

During strong changes in trend, with P-Red and P-Yellow, although red may be above yellow to start, I often see yellow climbing faster than red until it "pulls" red up further. When red turns up before yellow, it's often by several days. It tends to move up slower in this mode, not snap up.

I think that there are more patterns which can be exploited in the BNS indicators. For me, the P-indicators work as magnifiers of the relative movement. Sudden shifts in P-Yellow, P-Blue and P-Red are draw your attention when you look at them, but are less obvious to classify when using a mechanical system test.

Formulas

BNS-Chaikin:
-CO()

AvgPrice: (((H+L+C)/3)-Mov(C,35,S))/.015*Std((H+L+C)/3,35)
Alternatives to investigate:
(((H+L+C)/3)-Mov(C,35,S))/(.015*Std((H+L+C)/3,35))

StdDev:
Std(((H+L+C)/3),28)

RL-Proxy:
Stoch(8,5)

P-Blue:
1-Div(HHV(-CO(),28)+CO(),
.001+HHV(-CO(),28)-LLV(-CO(),28))

P-Yellow:
1-Div(HHV(Fml("AvgPrice"),28)-Fml("AvgPrice"), .001+HHV(Fml("AvgPrice"),28)-LLV(Fml("AvgPrice"),28))

P-Red:
1-Div(HHV(Fml("RL-Proxy"),28)-Fml("RL-Proxy"), .001+HHV(Fml("RL-Proxy"),28)-LLV(Fml("RL-Proxy"),28))

P-White:
Div(Fml("StdDev")-LLV(Fml("StdDev "),28),
HHV(Fml("StdDev "),28)-LLV(Fml("StdDev "),28))

----------
I played with a couple of other indicators to highlight strong changes in P-Yellow (and P-Blue) from low (and high) extremes.

Blue-Snap:
100*Max(0,(Ref(Fml("P-Blue"),-1)-Fml("P-Blue"))/ (If(Ref(Fml("P-Blue"),-1)>.9,.001,1-Ref(Fml("P-Blue"),-1))) )

Yellow-Snap:
100*Max(0, (Fml("P-Yellow")-Ref(Fml("P-Yellow"),-1))/ (If(Ref(Fml("P-Yellow"),-1)<.1,.001, Ref(Fml("P-Yellow"),-1))) )
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext