LSMA, Polynomial and other Regression with Experts and Indicators - page 2

 

LSMA does have it's uses

One of the early successful EAs that I wrote was a divergence EA using simple moving averages of 8/72. LSMA could give it an edge since the fast MA would be lots faster.

I've also used it in the last day to close a manual trade on GBPJPY in my live account for over 1000 pips.

Real Traders and Real Programmers can find a great deal of utility with it.

Ron(ForexMT4)

 

Robert,

Thank you very much for sharing your talent with us, I may represent many wannabe traders that get a lot of help from you, but we never express it, but to confront the negativity, I think people should know how many people stand in your side.

I can't contributed yet, but I will when I can, just like what you have been teaching us with your knowledge.

 

Glad you started the thread, and nice to have you here. The N4 TF indicator looks promising.

 

new N4 TF Indicator

robp:
Glad you started the thread, and nice to have you here. The N4 TF indicator looks promising.

I added a new version of the N4 indicator to the 3rd post.

It looks like I will need to zip these soon. I should have left a few more posts at the start.

Robert

 

MTF based on LSMA Colour

Robert,

Is it possible to create a MTF indicator base on the 3 colour LSMA rather than the usual one? The red / blue / yellow over 4 timeframes would be very useful.

Thanks and regards,

Paul.

 

LSMA DAILY EA Profit to date

FXDD Demo Account $3000 19th june 2007

GBPUSD

GPBJPY

EURCHF

Daily TF

closed profit +$136.25

current equity as at 12:24 pm GMT 29th june 2007 +$3478

Out of the box if i remember correctly

no manual interferance online 24/7

Robert,

I you have any suggestions / requirements for forward testing pairs /parameters let me know.

I have 2 pcs, can run upto 25 MT4 sessions and am fulltime on forex and EAs

Cheers

Damian

ps will post statement, settings, etc. at weekend

 

Ok, the basic idea of Linear Regression EA is to draw a relevant channel and trade within or breakout of it. So it can be divided on several tasks:

1. Calculate the all possible channels.

2. Select the most relevant channels. (I wrote channels because it's possible that there are more than one channel at the moment. Shorter channel can be within lager one and so on.)

3. Trade bounce or breakout.

1. All channels.

First we specify the range of channel searching. I think the best way here is using swings from higher TF's.(zigzag). But for the start it's better just take for instance 1000 bars, later we can optimize it. The end point is basically the most recent bar. Minimum length of a channel, say 20 bars. In that way we get 980 channels, each 1 bar shorter than the previous.

2. Selection of channels.

The hardest part. This is what I think should be done:

a. All divergent channels should be droped. Obviously, we are not interested in broken channels.

b. We get several clusters of unbroken channels. Now in each cluster we select a channel where standard deviation is minimum.

c. Now if channel is only one, that's great and we go to trading. But if there are several as usually happens and they point in different direction then we should somehow range them. I don't yet know how. Hope someone can help here.

3. Trading

When we have a channel we need to place trade, here we can use help of oscillators or something else.

I attached a screenshot taken in the TradeStation. Which shows how it can look. While channels are not perfect just for you to see the idea.

Comments and suggestions are welcome.

Wookey

Files:
linreg.jpg  104 kb
 

channels

wookey,

Thanks for the explanation.

Looks interesting. The idea for trading sounds similar to the Hoover method. The difference is determining which channel to use.

I have been meaning to write an EA for that method but have been too busy.

The idea is to use an MA cross within the channel. If the cross is in the center area of the channel no trade. If in the second level a trade is placed with first target either the center or opposite 2nd level. If outside the 2nd level a trade is placed with the same target ideas.

If this is added I will also include partial close at the targets and let the rest of the trade run to an opposite signal.

The EA at this time is very basic. I hope to be able to modify it to be profitable on lower timeframes. I have been looking at a few filter ideas like damiani_volatmeter and waddah_attar_explosion.

Any other ideas are welcome.

Robert

 

Damiani and WAE are good, but I've only seen them reliable at TF's lower than daily.

I think your NF LSMA is great. Is the plan to trade off of lower TF waves being filtered by each higher TF lsma?

MrPip:
wookey,

Thanks for the explanation.

Looks interesting. The idea for trading sounds similar to the Hoover method. The difference is determining which channel to use.

I have been meaning to write an EA for that method but have been too busy.

The idea is to use an MA cross within the channel. If the cross is in the center area of the channel no trade. If in the second level a trade is placed with first target either the center or opposite 2nd level. If outside the 2nd level a trade is placed with the same target ideas.

If this is added I will also include partial close at the targets and let the rest of the trade run to an opposite signal.

The EA at this time is very basic. I hope to be able to modify it to be profitable on lower timeframes. I have been looking at a few filter ideas like damiani_volatmeter and waddah_attar_explosion.

Any other ideas are welcome.

Robert
 

MTF_LSMA_in_Color

stav01:
Robert,

Is it possible to create a MTF indicator base on the 3 colour LSMA rather than the usual one? The red / blue / yellow over 4 timeframes would be very useful.

Thanks and regards,

Paul.

I thought of this as well. The problem is the limit on indicator buffers. For 3 colors we would be limited to 2 timeframes per indicator window.

I do have a version that plots one per window and have attached it to the third post.

I modified one that I have that also can use HMA, Ehlers iTrend and nonLagMA.

For those you need the 3 other indicators.

Robert

Reason: