Group Collaboration System Ideas

[SaidBitar]

SO quite over here :ohmy:

recently i was searching the net for trend following ideas and channel breakout to be specific, I came across one idea that is based on linear regression channel.
so instead of the the moving average the linear regression curve is used and the upper and lower channels are based on x ATRs away from it.

So i took the idea and added some touches and flavors to it and I could not make it a system that I will trade, but i managed to make it a mean reversion system.

The idea behind using the linear regression instead of moving average is to remove the lagging that exits in the moving average.

Anyhow here are the rules:
To stay aligned with the previous systems the following are not changed:
[ol]
[li]Index filter based on 50 days simple moving average[/li]
[li]minimum volume of 500K[/li]
[li]the close should be above the 100 days MA[/li]
[li]the close should be bellow the 5 days MA[/li]
[li]linear regression duration 20 days[/li]
[li]the bottom channel is 2.5* ATR(20) from the linear regression line[/li]
[li]Close bellow the Linear regression line[/li]
[li]Entry price is the lower channel[/li]
[li]Exit one profitable close (in the beginning I tested with close above the linear regression line)[/li]

[/ol]

[SaidBitar]

here are the results of single run from 1/1/2000 – 31/12/2015 on Russell 1000
20 positions 10% per position

[Nick Radge]

Quote:

SO quite over here

Everyone has been told to get on with their own systems!

[Anonymous]

looks my initial attempt is different to yours said:

[LeeDanello]

Darryl, I love your enhanced profit table

[SaidBitar]

Nick Radge wrote:

Quote:

SO quite over here

Everyone has been told to get on with their own systems!

Oops

[SaidBitar]

@DARRYL maybe it is the same I just made single run no mcs

[ScottMcNab]

Many thanks for posting the system Said….generous as always….

[Anonymous]

Maurice Petterlin wrote:

Darryl, I love your enhanced profit table

so do i maurice!

[JulianCohen]

Said would you mind posting your code for this as I’d like to see how you have used the regression line code please?

[Anonymous]

Julian Cohen wrote:

Said would you mind posting your code for this as I’d like to see how you have used the regression line code please?

hi julian i think its as easy as this:

Code:

linRegression = LinearReg(C,20);

[JulianCohen]

Doh! I thought it might have been the Clenow code Nick posted in the templates forum…I didn’t think to look at the AFL function list

[SaidBitar]

Code:

LongMAP = Param(“Long MA”,100,1,1000,1);
ShortMAP = Param(“Short MA”,5,1,1000,1);
LongMA = MA(C,LongMAP);
ShortMA = MA(C,ShortMAP);
//———————————————————————————
Cond1 = C > LongMA;
Cond2 = C < ShortMA; RegDura = Param("Linear Regression Duration",20,1,50,1); a = LinearReg(C,RegDura); b = a + 1 * ATR(RegDura); ATRMul = param("ATR Mult",2.5,1,10,0.5); g = a - ATRMul * ATR(RegDura); Cond3 = C < a; Cond4 = IndexFilt AND OptFilt AND HDBFilter ;

I used a as the linear regression line
b as the upper channel
and g as the lower channel

normally i use lousy names in the beginning then i make clean up if the system is ok

[Anonymous]

image resize test…

fail

[Anonymous]

hi said, do you resize your images before inserting? if so what size? all mine end up fuzzy

Said Bitar wrote:

here are the results of single run from 1/1/2000 – 31/12/2015 on Russell 1000
20 positions 10% per position

[Author]

Posts