I use linear regression channels and do very well. Like anything, it's all about how you use them.
http://www.elitetrader.com/vb/showthread.php?s=&threadid=54508
http://www.elitetrader.com/vb/showthread.php?s=&threadid=54508
Does anyone use linear regression to day trade the markets?
- Thread starter kevinmclark
- Start date
Not necessarily. Sometimes things are seldom discussed because a few people know how to profit from them and see no reason to share that particular knowledge. So questions in that area go unanswered except by the uninformed. But sometimes things go undiscussed because it's obvious to experienced techs that they are dead-ends. Like Carl Sagan said, "[Skeptics] laughed at Einstein, but they also laughed at Bozo the clown."
Nononsense did not mislead you. Regression, linear or nonlinear, is a form of modelling. If you really believe a security price can be modelled by a straight line, then by all means explore linear regression. But if this strikes you as silly and maybe even dangerous, then start looking elsewhere for your TA.
For some time I found the idea of LR interesting. I read an article in the last couple of years in Stocks and commodities about a LR based system, sorry don't remember which issue. I've experimented myself with LR using a number of periods, channel sizes and entry/exit rules but never found anything useful. However I've also become a big believer in different traders are going to "see" different things and have different outcomes even when looking at and using exactly the same tools/charts ect. Which simply means once you get a few ideas you'll have to test/try them for yourself.
TO THE PERSON WHO STARTED THIS THREAD:
I've read every response and haven't seen one that remotely knows what he's talking about.
LR channels absolutely do work.
Price data, like all data contains an inherent linear regression line. Price HAS to react to the LR line and its parallels. HAS to . . . has no choice.
Note . . . I'm talking here about the Drawing Tool called linear regression channel . . . not the computation that looks like a moving average. There's little standardization on names so you have to be careful. Hopefully, I'm talking about the one you mean.
Try this.
Start by drawing the LR Line by itself. Then add 2 std dev channels. Erase those and build a "Raff Channel" based on the greatest width of the Close from the LR line. Most price action is contained by one of those two types of channels. You can also experiment with 1 std dev, 4 std dev and channels based on other highs and lows. (You may want to study Dr Andrews work on Action/reaction Lines to understand how to do this.)
Once you get the channel types down and you know what each one can tell you, start applying them to the various swings.
I won't reveal the rest because it's proprietary, but I'll say this:
The key to the proper application of LR Channels is knowing how much data to include in the channel. When does it start . . . when does it end? Most people (99%) just apply it to the swing in front of them and bitch because "it doesn't work."
One swing . . . three swings . . . two days . . . a week . . . experiment. You can figure it out.
How much data to use . . . that's where the answer lies.
I've read every response and haven't seen one that remotely knows what he's talking about.
LR channels absolutely do work.
Price data, like all data contains an inherent linear regression line. Price HAS to react to the LR line and its parallels. HAS to . . . has no choice.
Note . . . I'm talking here about the Drawing Tool called linear regression channel . . . not the computation that looks like a moving average. There's little standardization on names so you have to be careful. Hopefully, I'm talking about the one you mean.
Try this.
Start by drawing the LR Line by itself. Then add 2 std dev channels. Erase those and build a "Raff Channel" based on the greatest width of the Close from the LR line. Most price action is contained by one of those two types of channels. You can also experiment with 1 std dev, 4 std dev and channels based on other highs and lows. (You may want to study Dr Andrews work on Action/reaction Lines to understand how to do this.)
Once you get the channel types down and you know what each one can tell you, start applying them to the various swings.
I won't reveal the rest because it's proprietary, but I'll say this:
The key to the proper application of LR Channels is knowing how much data to include in the channel. When does it start . . . when does it end? Most people (99%) just apply it to the swing in front of them and bitch because "it doesn't work."
One swing . . . three swings . . . two days . . . a week . . . experiment. You can figure it out.
How much data to use . . . that's where the answer lies.
I don't post much anymore as my health is still shaky. But reading this post from my hospital bed I can tell you one thing: Our company consistently uses a LR based system with a very high degree of success. As Jack says, you have to find the key on how to use it for your self. No one will share their working system.
Cheers.
Cheers.
(And here for all this time some of us have figured you've been totally immersed with Darren Clifford in learning the intracacies of pair trading, Maestro...
Here's hoping you successfully take care of whatever has prompted your trading world partial time-out!)
Here's hoping you successfully take care of whatever has prompted your trading world partial time-out!)
I was afraid for your health during your absence. I regret hearing that your health is still shaky. Feel and get better soon. We will pray for you.
Linear regression is one of the most powerful tools that MAESTRO showed me how to use. Since this time, a period of several months, i have been able to use his work to my advantage. I am forever grateful.
It's not only what you use, it's how you use it. There lies the difference.
Please get better Maestro. You are deeply missed.