|
|
Perhaps the above is enough for many traders, but a few
further MetaStock nuances can add to the value of the
information you've uncovered. For example, wouldn't you like
to know which stocks have met the chosen crossover criteria in
the past, say, five days? And wouldn't it be handy to be
able to sort your newly discovered stocks in order of price or
volume? If so, read on for a few more simple tips.
1. Go back to the main Explorer tool section, highlight your
"Moving Average Crossover" Exploration, and hit the "edit" key
this time. You can now make alterations to your Exploration.
Ignore the upper "Notes" section and click on Column A first.
You will see a large white field for entry of formulas and a
small field in the lower left, entitled "Col Name." Simply put a
"c" in the large formula section and "Close" in the column name
section. Repeat these actions for Column B, with "v" and
"Volume" respectively. Now when your Exploration presents you
with your data, you can easily sort by price (c) or volume (v).
2. Finally, click on the "Filter" tab again to slightly
modify your Exploration formula. The way you have it set up
initially tells MetaStock to find all stocks which meet the
criteria today. You now want it to find all stocks that have met
these criteria over the past five days. The answer is the
MetaStock Alert function, which is written "Alert( A , Number )
where "A" is any formula you care to choose, and "Number" is the
number of days. So now you put your original formula in the
place of A. The result is: "Alert( Cross( Mov(C,3,E) ,
Mov(C,10,E) ) ,5)" without the quotation marks. Save your new
Exploration with the "OK" button and you're ready to find all
stocks whose 3 day moving average passed above the 10 day moving
average in the past five trading days!
The above information should allow you to write further
Explorations by simply changing the numbers. If you prefer to
use Exponential Moving Averages instead of Simple Moving
Averages, change "s" to "e" in the formulas. You can also open
up the ready made Equis Explorations, investigate how they're
written, and change them with the "Edit" command (then saving
with a new name). A further step is to investigate the hundreds
of formulas available here on this web site and modify them in
the same way. This is the quick and easy way to learn how to
program with MetaStock. Follow the examples given by all the
kind and clever MetaStock users who have gone before you, and
tweak, tweak, tweak.
(Go
Top) |
|
The Barnes' Acceleration measures rate of price change as
opposed to price
levels
If the Barnes' Acceleration sustains the value of -1 for many
days then the
security may be ready to show strong trend or it may already be
trending.
Examine the chart for prolonged values at -1. This may indicate
an upcoming
stall or turnaround. The number of days needed may be different
depending on
the type of issue. A utility stock may need to sustain the -1
level for 10
days whereas a highly volatile technology stock may need to
sustain the -1
trend for as little as 5 days.
From the 1981 Technical Commodity Yearbook, Robert M. Barnes
formula 1: if(mov(fml("Barnes' acceleration",2) -
ref(fml("Barnes'
acceleration",2),-1),20,e)>0.0001,1, if(mov(fml("Barnes'
acceleration",2) -
ref(fml("Barnes'
acceleration",2),-1),20,e)<-0.0001,-1,0))
formula 2: mov((c-ref(c,-1))/ref(c,-1),daysm,e)
(Go
Top) |
|
Based on the premise that closing price may be predictable
based on previous
closes
See (1981 Technical Commodity Yearbook Robert M. Barnes Van
Nostrand
Reinhold 1981) for theory and applications.
formula 1: if(fml("Barnes' adaptive
forecast",2)>0.05,1,if(fml("Barnes'
adaptive forecast",2)<-0.05,-1,0))
formula 2: mov(c,dayf,e) - ref(mov(c,dayf,e),-1)
(Go
Top) |
|
See (1981 Technical Commodity Yearbook Robert M. Barnes Van
Nostrand
Reinhold 1981) for theory and applications.
if (c - mov( c, dayf, e) > pf,
{ Then Action } 1,
{ Else Action } if ( mov( c, dayf, e) - c > pf, { Then } -1, {
Else } 0))
{ Notice that comments may be placed within braces }
(Go
Top) |
|
cum ((if ((mov((C-ref(C,-1)),21,s))>0,1,-1) *
pwr(((mov((pwr(C-ref(C,-1),2)),21,s))+1),.5)) +
((pwr(((pwr(C-ref(C,-1),2))+1),.5))) * if ((C>ref(C,-1)),1,-1))
(fml(" BPDL Trend Filter") - (ref((fml(" BPDL Trend
Filter")),-21))) / ((hhv(fml(" BPDL Trend Filter"),21)) -
(llv(fml("BPDL Trend Filter"),21)))
BUY -1 SELL 1
(Go
Top) |
|
Col A: CLOSE
Filter BarsSince(EngulfingBull())<=5 AND
BarsSince(ROC(C,60,%)<-15)<=5 ANDBarsSince(Stoch(9,1)<10)<=5
Filter enabled Yes
Periodicity Daily
Records required 1300
(Go
Top) |
|
enter long
When(Mov( Mid(C, opt1) ,opt1,E),>,Mov(Mid(C, opt1),opt2,E))AND
When(Ref(Mov(Mid(C,opt1),opt1,E),-1), <=
,(Ref(Mov((Mid(C,opt1)),opt2,E),-1)))AND
When(Mov(Abs((Mo(opt3))),opt4,E),>,Ref(Mov(Abs((Mo(opt3))),opt4,E),-1))
enter short
When(Mov( Mid(C, opt1) ,opt1,E),<,Mov(Mid(C, opt1),opt2,E))AND
When(Ref(Mov(Mid(C,opt1),opt1,E),-1), >=
,(Ref(Mov((Mid(C,opt1)),opt2,E),-1)))AND
When(Mov(Abs((Mo(opt3))),opt4,E),>,Ref(Mov(Abs((Mo(opt3))),opt4,E),-1))
OPT 1: 5 to 20 step 1
OPT 2:10 to 16 step1
OPT3:5 to 15 step 1
OPT4:20 to 29 step 1
but you are free to change any value of OPT!
(Go
Top) |
|
Col A: CLOSE
Col B: ROC(C,5,%)
Filter: (ROC(C,5,%)>10 OR ROC(C,5,%)<-10) AND C>5
Filter enabled: Yes
(Go
Top) |
|
{created by Jim Greening}
The basic idea behind a MetaStock binary wave is to use "if"
statements on several MetaStock indicators and have them return
plus one for a bullish indication, minus one for a bearish
indication, and zero for a neutral condition. Then you add them
all up for your binary wave indicator. I decided to format all
my indicators so they could be plotted as a histogram. For these
indicators plotting as histograms, positive is bullish and
negative is bearish. To cut down on whipsaws, I decided that
over +5 would be bullish, under -13 would be bearish and
anything in between would be neutral. Therefore my binary wave
formulas are:
BW2 Demand Index
If(Tema(DI(),21) > 5,+1,If(Tema(DI(),21) < -13,-1,0))
BW3 Linear Regression Slope
If(Tema(10000*LinRegSlope(C,34)/C,34) > 5,+1,
If(Tema(10000*LinRegSlope(C,34)/C,34) < -13,-1,0))
BW4 CCI
If(Tema(CCI(21),21) > 5,+1, If(Tema(CCI(21),21) < -13,-
1,0))
BW5 ROC
If(Tema(ROC(C,21,%),21) > 2,+1,If(Tema(ROC(C,21,%),21) <
-2,-1,0))
BW6 Money Flow
If(Tema(MFI(21),21)-50 > 5,+1,If(Tema(MFI(21),21)-50
< -5,-1,0))
BW7 CMO
If(Tema(CMO(C,21),21) > 5,+1,If(Tema(CMO(C,21),21)
< -5,-1,0))
BW8 VAR ma
If(Mov(C,21,VAR) > Mov(C,55,VAR) AND
HHV(Mov(C,233,VAR),5) =
HHV(Mov(C,233,VAR),13),+1,If(Mov(C,21,VAR) <
Mov(C,55,VAR) AND LLV(Mov(C,233,VAR),5) =
LLV(Mov(C,233,VAR),13),-1,0))
The next formula just adds up the binary wave.
BW Add
Fml("BW2") + Fml("BW3") + Fml("BW4") + Fml("BW5") +
Fml("BW6") + Fml("BW7") + Fml("BW8")
Next, I decided to do something a little different. Since the
whole purpose of this test is to catch a trending stock, I
decided to add an amplifier that would get larger as the trend
got stronger. Since I like Fibonacci numbers, I decided to use
Rsquared as a measure of trend strength and base my amplifier on
Fibonacci numbers. The formula I finally came up with after a
lot of tinkering follows.
BW Amplifier
If(RSquared(C,21) > 0.8,5,If(RSquared(C,21) >
0.6,3,If(RSquared(C,21) > 0.4,2,
If(RSquared(C,21)>0.2,1,0.5))))
The last step in constructing the binary wave was to decide on
the smoothing and put it all together. Of course, I used tema
smoothing.
Tema Binary Wave Composite
Periods := Input("Enter Tema Smoothing
Periods",8,233,21);
Tema(Fml("BW Add")*Fml("BW Amplifier"),Periods)
The final step is to come up with a system test for the Tema
Binary Wave Composite. Remember, the binary wave is just made up
of a bunch of technical indicators that I give a +1 value when
bullish, 0 when neutral, and -1 when bearish. Then they are
summed and smoothed. So in general a positive value is bullish
and a negative value is bearish. Also a rising number is bullish
and a falling number is bearish. Therefore you could use a zero
crossover to the upside as a buy signal and a crossover to the
downside as a sell signal. If you had a good algorithm, you
could also use a rise from a negative peak (or trough) as a buy
signal and a fall from a positive peak as a sell signal. I
decided to use a 8 day moving average of the BW with a crossover
of the BW for my algorithm in an attempt to get an early signal
on a rise from a negative peak. It does have the disadvantage of
finding way too many peaks so I only use it as an Alert. For
confirmation I use the QStick function and a variable moving
average function.
QStick was developed by Chande as a way to quantify
candlesticks. Since the difference between the open and close
prices lies at the heart of candlestick charting, QStick is
simply a moving average of that difference. Negative values of
QStick correlate to black candlesticks, positive values to white
candlesticks. Since in general black candles are bearish and
white candles are bullish, this indicator can also be plotted as
a histogram and interpreted the same was as the Binary Wave. The
formula is:
Periods := Input("Enter Periods",1,233,34);
Tema(Qstick(Periods),Periods)
Now to get my open long signal I use the ALERT signal with an 8
day vma BW crossover of the BW. Then to actually get the signal,
I have to have both the QStick rising and the 21 day vma greater
then the 55 day vma.
Therefore my buy signal became:
Enter Long
Alert(Cross(Fml("Tema Binary Wave Comp"),
Mov(Fml("Tema Binary Wave Comp"),8,S)),21) AND
HHV(Tema(Qstick(34),34),5) = HHV(Tema(Qstick(34),34),13) AND
Mov(H,21,VAR) > Mov(H,55,VAR)
Since the market has an upward bias, I wanted my sell signal to
be more restrictive. Therefore instead of trying to catch a fall
from a positive peak as my sell alert, I wanted a crossover of
an optimized negative number. I still used QStick and vma to
confirm and also added that the close should be lower than
yesterdays low.
Therefore, my sell signal became:
Enter Short
Alert(Cross(-opt2,Fml("Tema Binary Wave Comp")),8) AND
Tema(Qstick(34),34) < -0.1 AND
C < Ref(L,-1) AND
Mov(L,21,VAR) < Mov(L,55,VAR)
Then I wanted exit conditions that were less then full signal
reversals. I decided that the BW being less than a negative
number would be my primary close long signal, but I also wanted
confirmation from other indicators. After a lot of trial and
error I used the following:
Close Long
Fml("Tema Binary Wave Comp") < -opt1 AND
Tema(Qstick(34),34) < 0 AND
LLV(Mov(L,21,VAR),5) = LLV(Mov(L,21,VAR),13)
Close Short
Fml("Tema Binary Wave Comp") > 0 AND
Tema(Qstick(34),34) > 0.08
Finally I also used Fibonacci numbers for my optimization:
Opt 1: Min 3, Max 13, Step 5
Opt 2: Min 3, Max 13, Step 5
(Go
Top) |
|
{I was reading in Perry Kaufman's latest book and he
described a little
oscillator he called "Body Momentum". This simply calculates the
momentum of
the closes above the opens versus the closes below the opens.
The theory is
that as prices move up, closing prices will be higher than
opening prices
and vice-versa for down. If this oscillator is above 70 then the
whites
(Candle-sticks) dominate and below 30 the blacks are dominant.}
{I also added a 3 day moving average to the calculation (for
smoothing).}
{Here is the code:}
Lb:=Input("Look-Back Period?",3,60,14);
B:=CLOSE - OPEN;
Bup:= Sum(B > 0, Lb);
Bdn:= Sum(B < 0, Lb);
BM:=(Bup/(Bup+Bdn))*100;
Mov(Bm,3,S)
(Go
Top) |
|
From: Ton Maas
According to most analysts, the Chaikin Oscillator, a diverse
accumulation/distribution line, is a very good alternative to
the OBV (On Balance Volume) indicator. Chaikin Oscillator basics
are that a healthy trend will be confirmed by a healthy,
positive volume development in the trend direction. The MFI
(Money Flow Index) can also substitute for the Chaikin
Oscillator.
Chaikin Oscillator formula:
Mov(cum(((C-L)-(H-C)/(H-L))*V),3,E)-Mov(cum(((C-L)-(H-C)/(H-L))*V),10,E)
(Go
Top) |
|
Recently, the "group" was able to supply me with the formula
for making a Histogram out of the "bands". I find this the most
useful application of Bollinger's formula. The following is the
picture I draw:
((C+2*Std(C,20) - Mov(C,20,S)) / (4*Std(C,20)))*4 - 2
Under "properties", I then drop in +2 and -2 (because I'm not
bright enough to program them in permanently). I think this is a
much better view of the bands. As the price moves up and down as
a % of the band width, all the classic applications of other
"oscillator type" indicators work well (divergence,
support/resistance, and overbought/oversold conditions when the
price exceeds the Standard Dev. of +/-2).
This is just one of ten indicators that I use ... but, for
traders trying to understand Bollinger's "envelopes", I think
this reconfiguration gives a simpler, cleaner view which allows
the technician to analyse the underlying issue without the
"squiggles".
(Go
Top) |
|
I use the following indicators to show the price reversal of
Bollinger Band penetration:
Name: Upper BB Hookdown
Formula:
UpperBB:= Mov(C,20,S) +(2*(Std(C,20)));
C < UpperBB AND Ref(C,-1) > Ref(UpperBB,-1);
Name: Lower BB Hookup
Formula:
LowerBB:= Mov(C,20,S) -(2*(Std(C,20)));
C > LowerBB AND Ref(C,-1) < Ref(LowerBB,-1);
(Go
Top) |
|
I am sure Steve has done something better, but here is a
simple (MetaStock) formula allowing you to draw Bollinger Bands
as an oscillator:
100*(C-Mov(C,20,S)+2*Stdev(C,20))/(4*Stdev(C,20))
(Go
Top) |
|
enter long
high>(mov(Close,20,S)-std(Close,20,2)) and
ref(low,-7)<ref((mov(Close,20,S)-std(Close,20,2)),-7)
exit long
close<(mov(Close,20,S)+std(Close,20,2)) and
ref(close,-7)>ref((mov(Close,20,S)+std(Close,20,2)),-7)
and
Mov((RSI(14)- LLV(RSI(14),14))
/(HHV(RSI(14),14)-(LLV(RSI(14),14))),14,E)*100<70 and
ref((Mov((RSI(14)- LLV(RSI(14),14))
/(HHV(RSI(14),14)-(LLV(RSI(14),14))),14,E)*100),-3)>70
and
(mov(Close,20,S)+std(Close,20,2))>(mov(c,89,s)+(.062*(mov(c,89,s))))
(Go
Top) |
|
John Bollinger describes BWI (Band Width Indicator) as the
width of the bands divided by the average of the price:
4*(std(C,20))/mov(C,20,S)
I don't know if adding the moving average changes the usefulness
of the prospecting; anyway, this is what Bollinger is
suggesting.
I have written a MetaStock exploration to spot stocks whose BWI
has reached extreme low readings. This shows when the BWI is at
lower than its highest level for the last 250 days, divided by
3:
hhv(4*(std(C,20))/mov(C,20,S),250)/3
The stocks that pass this screening are usually in a
non-trending mood, or rather in an
horizontal trend where the Bollinger Bands normally represent
support and resistance levels. Otherwise, there are cases where
the stock is just pausing before resuming a trend. In this
second case the BWI doesn't remain under the trigger level for a
long time.
A further remark is that when the stock enters a low-BWI period,
it is often retesting a
previous support or resistance level.
Although I think BWI extreme lows are an interesting way to find
low risk / low volatility stocks, they don't give any clue as of
the direction of the following move.
(Go
Top) |
|
From: Philip Schmitz
MetaStock v6 does not appear to provide an indicator which shows
the width of Bollinger
Bands, so I have concocted a simple one to suit my own needs:
"Band Width" = BBandTop(C, 70, E , 2) - BBandBot(C, 70, E , 2)
As a next step, I would like to devise an indicator which tells
me how the current value of
"Band Width" relates to the overall range of Band Widths for a
specified period, or, since my interest is commodities, the life
of the contract -- in other words all data loaded. Where, on a
percentage basis, does it fall?
(Go
Top) |
|
BOSS -- Synergy with Bollinger by John Lowe (March 1998 issue
of TAM, a Dutch
TA mag)
In this article John Bollinger gets mentioned as insisting on
using a Price/Close
indicator in conjunction with a combined Price/Volume indicator.
For example, Price
as a moving or exponential average, the Typical
Price(High+Low+Close/3) or one of
the other on this theme of existing varieties. Bollinger strives
for synergy, which has
to be confirmed by two of three indicators based on:
Closing-price, price and volume, the Bollinger Optimised Synergy
System (BOSS):
1st criteria -- Bollinger Bands are best used in conjunction
with Wilders' RSI(9 or 14),
an indicator based on closing price.
2nd criteria -- Price and volume, combined in the Chaikin
Oscillator, are the other part
of the BOSS.
According to most analysts, the Chaikin Oscillator, a diverse
accumulation/distribution line, is a very good alternative to
the OBV indicator.
Chaikin Oscillators' basics are that a healthy trend will be
confirmed by a healthy,
positive volume-development in the trend-direction. The Chaikin
Oscillator can be
substituted for with the Money Flow Index (MFI).
Chaikin Oscillator formula:
Mov(cum(((C-L)-(H-C)/(H-L))*V),3,E)-Mov(cum(((C-L)-(H-C)/(H-L))*V),10,E)
(Go
Top) |
|
A:Close
B:{Signal Buy=-1,Sell=1}
If(ADX(14)>30 and PDI(14)>MDI(14),-1,If(ADX(14)>30
and PDI(14)<MDI(14),1,0))
C:{setup}Ref(H,-2)>=Ref(H,-1) and Ref(H,-1)>=H and
Ref(L,-2)<=Ref(L,-1) and Ref(L,-1)<=L
D:{Entry}If(ADX(14)>30 and PDI(14)>MDI(14) and
Ref(H,-2)>=Ref(H,-1) and Ref(H,-1)>=H and
Ref(L,-2)<=Ref(L,-1) and
Ref(L,-1)<=L,HHV(H,3)+.125,IF(ADX(14)>30 and
PDI(14)<MDI(14) and Ref(H,-2)>=Ref(H,-1) and
Ref(H,-1)>=H and Ref(L,-2)<=Ref(L,-1) and
Ref(L,-1)<=L,LLV(L,3)-.125,0))
E:{Stop}If(ADX(14)>30 and PDI(14)>MDI(14) and
Ref(H,-2)>=Ref(H,-1) and Ref(H,-1)>=H and
Ref(L,-2)<=Ref(L,-1) and
Ref(L,-1)<=L,LLV(L,3)-.125,IF(ADX(14)>30 and
PDI(14)<MDI(14) and Ref(H,-2)>=Ref(H,-1) and
Ref(H,-1)>=H and Ref(L,-2)<=Ref(L,-1) and
Ref(L,-1)<=L,HHV(H,3)+.125,0))
F:ADX(14){Higher the better}
Filter:ColB and ColC
(Go
Top) |
|
Boomers buysig
enter long
((adx(14)+adx(27))/2)>30 and pdi(27)>mdi(27)
exit long
c<=prev(llv(c,15)-.5, 1) or c<=.75*hhv(c,10)
Boomers watchsig**
enter long
prev(h,1)<=prev(h,2) and prev(l,1)>=prev(l,2) and
BullHarami()
exit long
c<=prev(llv(c,15)-.5, 1) or c<=.75*hhv(c,10)
Boomers watchsig 2**
(Ref not prev)
enter long
ref(h,-1)<=ref(h,-2) and ref(l,-1)>=ref(l,-2) and
BullHarami()
exit long
c<=ref(llv(c,15)-.5,-1) or c<=.75*hhv(c,10)
(Go
Top) |
|
These are a collection of bottom signals. The search returns
1 for Ok and 0 for not ok.
- CLOSE
- EngulfingBull()
- MorningDojiStar()
- MorningStar()
- WhiteSoldiers()
(Go
Top) |
|
BradCCI: From Bill S.
Plot 1: BradCCI Line 1:
(((H+L+C)/3)-Mov(C,28,S))/(.015*Std(C,28))
Plot 2: BradCCI Line 2: Std(((h+l+c)/3),28)
To Line 1, you can also add trend lines, if you wish:
Plot 1:
1. BradCCI Line 1: (((H+L+C)/3)-Mov(C,28,S))/(.015*Std(C,28))
2. trend(100,100)
3. trend(-100,-100)
4. trend(0,0)
(Go
Top) |
|
Name: RSI derivative index (EL) - C. Brown {EasyLanguage}
Formula:
Base:=Mov(RSI(14),6,S);
ATRcustom:=HHV(
ValueWhen(1,RSI(14)>Ref(RSI(14),-1),ATR(1))OR
ValueWhen(2,RSI(14)>Ref(RSI(14),-1),ATR(1))OR
ValueWhen(3,RSI(14)>Ref(RSI(14),-1),ATR(1)),14);
Part1:=
{up coefficient is a factor 2.3 and added}
(2.3*(Mov(ATRcustom, 15,S)));
Part2:=
{down coefficient is factor 2.1 and subtracted}
(2.1*(Mov(ATRcustom, 15,S)));
{Part3=if RSI closes up or equal to, than part1,
else, if RSI closes down, part2}
If( Base>=Ref(Base,-1),
{RSI up=TRUE, then} Base+Part1,{else}
{RSI up=FALSE,then} Base-Part2)
(Go
Top) |
|
The system is a trend follower that appears to get you in at
the early in a trend. If the trend breaks down for any reason,
the system seems to take you out with relatively little pain,
and there is a relatively high
percentage of losing trades (usually around 50%). Therefore, the
system seems to perform best on issues that are prone to make
prolonged moves. The trick is to find those issues. I do admit
that the system is not perfect;
for instance, it is my belief that the exit could be improved on
winners to preserve more profit. However, I've been unable to
develop an alternative exit that improves the system return.
I've been trading this system myself for about a year and have
had good results. Even in the April-September period when
everything seemed to stall and move sideways, I was, at least
able to hold my own and maintain my capital until the October
break-always started to occur. For awhile, until I got bored
with it, I phantom traded this system in the Yahoo Investment
Challenge. I typically made about 20% a month using the system
in that venue.
Buy
n :=opt2{Time periods};
BullFear := (HHV(HIGH,n) - LLV(HIGH,n))/2 + LLV(HIGH,n);
Cross(CLOSE,bullfear) AND
DX(10) > opt1
Sell
n :=opt2{Time periods};
BearFear := (HHV(LOW,n) - LLV(LOW,n))/2 + LLV(LOW,n);
CLOSE < bearfear
Optimize the time periods from 10 to 50 in increments of 1 while
testing the DX from 5 to 30 in increments of 5 (you can do it in
increments of 1 but it takes longer). Once the Optimal time
period is determined in this manner,
then retest with the determined optimal time period and the DX
in increments of 1. Note that this system is intended to be a
stop and reverse system and you can use it to go short as well
if you'd like to.
(Go
Top) |
|
enter long:
n :=opt2{Time periods};
BullFear := (HHV(HIGH,n) - LLV(HIGH,n))/2 + LLV(HIGH,n);
Cross(CLOSE,bullfear) AND
DX(10) > opt1
close long:
n :=opt2{Time periods};
BearFear := (HHV(LOW,n) - LLV(LOW,n))/2 + LLV(LOW,n);
CLOSE < bearfear
(Go
Top) |
|
Here's an excellent short article from Jim Greening, showing
how MetaStock system tests can be built up . . .
This week I'm going to discuss my third MetaStock Profit System
Test - 03_Tema PDI - MDI, ADX (Vol Required). This test is based
on Wilder's directional movement indicators. As the MetaStock
manual indicates, Wilder says a buy signal occurs when PDI - MDI
moves above zero and a sell signal occurs when PDI-MDI falls
below zero. I started with that thought and experimented a
little. Wilder used 14 day periods to calculate his PDI and MDI
functions. Since I like Fibonacci numbers, I used 13 days
instead. Also I like to smooth my indicators so I used Tema
smoothing. My custom PDI - MDI formula then became:
Tema PDI - MDI
Periods := Input("Enter Tema Smoothing Periods",8,55,13);
Tema(PDI(13) - MDI(13),Periods)
I started with the idea that I would take the PDI-MDI crossover
of an optimized number as my basic buy and sell trigger.
However, this number did not have to be zero and did not have to
be the same for entering long and entering short. After a lot of
trial an error I decide -1, -3, or -5 would be my enter long
number and -5, -13, or -21 would be my enter short number. This
makes sense since the market is biased to the up side, so
entering a little under zero would get us in a little earlier.
Also down moves tend to be fast an extreme and this would only
let us in short for larger, faster down moves which is what I
wanted. Finally I wanted some way to reduce the number of false
signals and I wanted to do that with directional movement
indicators only so this test would be completely uncorrelated
with my other tests.
For long positions, I notice that most up moves started when adx
was low and that adx climbed during the move to a max and then
started to fall at the end of the move. Therefore, I thought an
adx max and min for a buy signal would help reduce false
signals. After some experimenting, I set the min at 8 and the
max at 21. I also noticed that most good buy points occurred
when MDI and ADX were close together so I decided that the
difference between the two should be small. After more
experimenting, I decided on the following for my open long
signal:
Open Long:
Alert(Cross(Fml("Tema PDI - MDI"),opt1),13) AND
MDI(13) - ADX(13) <= 4 AND
MDI(13) - ADX(13) >= -2 AND
ADX(13) >= 8 AND
ADX(13) <= 21
To close my open long position I wanted the PDI-MDI to be less
than opt1. When a stock starts to drop, the MDI starts to rise,
so I wanted the MDI to be greater than a certain number to close
a position. Finally, since markets are biased upwards, I also
wanted the 55 day variable moving average to be dropping before
I closed the position. Therefore, the close long became:
Close Long:
Fml("Tema PDI - MDI") < opt1 AND
MDI(13) > 21 AND
LLV(Mov(L,55,VAR),5) = LLV(Mov(L,55,VAR),13)
To open a short position, I wanted the PDI-MDI to cross below a
fairly high negative number. I wanted confirmation in that the
adx was still fairly high when that happened. The answer was:
Open Short:
Alert(Cross(opt2,Fml("Tema PDI - MDI")),8) AND
ADX(13) > 34
To close the short position, I only wanted PDI-MDI to be greater
than a certain positive number. I don't like a lot of
confirmations for closing shorts. With the bias being up, you
need to close shorts fast. My close Short and optimization
became:
Close Short:
Fml("Tema PDI - MDI") > 13
Optimization:
Opt1: Min = -1 Max = -5 Step = 2
Opt2: Min = -21 Max = -5 Step = 8
That's it. Any comments or questions?
(Go
Top) |
|
ColA: CLOSE
Filter BarsSince(EngulfingBear())<=5 AND
BarsSince(ROC(C,60,%)>15)<=5 AND BarsSince(Stoch(9,1)>90)<=5
Filter enabled Yes
Periodicity Daily
Records required 1300
(Go
Top) |
|
The Breadth Thrust indicator is a market
momentum indicator developed by Dr. Martin Zweig. The Breadth
Thrust is calculated by taking a 10-day exponential moving
average of the advancing issues divided by the advancing plus
declining issues.
According to Dr. Zweig a Breadth Thrust
occurs when, during a 10-day period, the Breadth Thrust
indicator rises from below 40 percent to above 61.5 percent. A
"Thrust" indicates that the stock market has rapidly changed
from an oversold condition to one of strength, but has not yet
become overbought.
Dr. Zweig also points out that there have
only been 14 Breadth Thrusts since 1945. The average gain
following these 14 Thrusts was 24.6 percent in an average time
frame of 11 months. Dr. Zweig also points out that most bull
markets begin with a Breadth Thrust.
To plot the Market Breadth in
MetaStock™ for Windows you will need to:
-
Create a composite security of the
Advancing Issues + Declining Issues in The DownLoader™.
-
In MetaStock open a chart of the
composite and a chart of the Advancing Issues.
-
Tile the charts so you can see both of
them on the screen.
-
Drag the plot of the composite into the
chart of the Advancing Issues.
-
Create the custom indicator: mov( C / P,
10, E ), then plot it on top of the plot of the composite
(the composite's plot will turn a purplish color ). If you
get a flat line then it was not plotted directly on top of
the composite's plot.
-
You can then right-click on the Breadth
Thrust, select Breadth Thrust Properties, go to the
Horizontal Lines page and add horizontal lines at 40 and 60.
(Go
Top) |
|