Thanks for the suggestions and encouragement (Wint, Diamondtrim, MAESTRO, JR97, & Cluseau). I have been using a linear regression line with standard error channels with some success; however, Iâm still learning how to best utilize this tool. Like JR97 I monitor multiple timeframes for potential targets and support & resistance levels. Additionally, itâs helps me to better interpret price action (defined as LHs, HLs, DTs, & DBs).
Does anyone use linear regression to day trade the markets?
- Thread starter kevinmclark
- Start date
I am very familiar with modeling data using linear regression and the mathematics behind it and its limitations. This is something youll have to study in school to get a firm grasp. Unfortunately, I cant help you with linear regression lines. Never heard of them outside of technical analysis. I suspect they are very different tools.
I have not regressed daytrading data though if it would be a very powerful tool to test trading ideas. It doesnt create trade ideas in itself rather its more of a statistical/mathematical tool.
Hope this helps.
I have not regressed daytrading data though if it would be a very powerful tool to test trading ideas. It doesnt create trade ideas in itself rather its more of a statistical/mathematical tool.
Hope this helps.
take a look at the eSignal intrepretation of Linear Regression on the charts, oh, not the advanced but regular or basic charts.
it really is a very powerful trading technique...
As regression is a first order method in statistics, it has been and is applied in MANY fields, sometimes with great success.
In speculating, a good way of looking at it is to say that linear regression is one of the popular ingredients in many "indicator" recipies. Some will make good money for the few; few make any money for the many.
nono
In speculating, a good way of looking at it is to say that linear regression is one of the popular ingredients in many "indicator" recipies. Some will make good money for the few; few make any money for the many.
nono
I have done a lot of calculations on regression channels and moving averages in the past. Found something that worked a few years ago using moving average, but never found anything using regression. Sure I found some stocks it worked on for a period of time, but never that was consistent in a sector or stocks in same risk class. If you find anything good feel free to post it though
ORM
Stick with the basic simple stuff.
The days pivot points have worked for the locals in the pits for ages.
When was the last time you saw a pit trader lug a IBM workstation with 6 Xeon chips at 3.6gh and 6 15K scsi hard drives set up in a raid array with 6 gigs of ram and 8 19" screens into the pit?...
Basic support and resistence, maybe moving average crosses and bingo you should be good to go.
Now remember your 3 "Ms" Mind, Money management, Mental discipline (Mix, Match, Massage those 3 Ms as you see fit.)
The days pivot points have worked for the locals in the pits for ages.
When was the last time you saw a pit trader lug a IBM workstation with 6 Xeon chips at 3.6gh and 6 15K scsi hard drives set up in a raid array with 6 gigs of ram and 8 19" screens into the pit?...
Basic support and resistence, maybe moving average crosses and bingo you should be good to go.
Now remember your 3 "Ms" Mind, Money management, Mental discipline (Mix, Match, Massage those 3 Ms as you see fit.)
There are always people who say "XYZ doesn't work" and for them nothing usually does. EVERYTHING works . . . some things better than others . . . but you can make anything work.
The real issue is what you're trying to do.
If you are trying to develop a methodology that will help you understand the market and be able to make a highly accurate prediction at any point, then you're in for a lot of hard work. There's a great chance that you don't have the know-how, IQ, funding or tools to succeed. You're going to become a statistic.
If, on the other hand, you are TRYING TO MAKE MONEY . . . it doesn't take a lot to do it. Develop a simple method that will earn you $100/day. When you have your hundred bucks, quit for the day. Go do something else. Do that every day. Increase contracts slowly, say one a month over 5-6 months.
By setting yourself up with simple, low expectations, you stay away from all but the highest probability plays. (Of course, this assumes you have the patience and self-discipline to wait.)
Here's an example of a simple way to do this with linear regression:
Wait for the 9 pd ema to be above the 18 pd ema on the 15 min chart. Once it is, look for the same alignment on the 3 min chart. Wait for a pullback. Draw the LR channel. Enter on a breakout of the LR channel. (Don't like the 9-18? Use the 5-35, 5-15, 5-20, or whatever suits you.)
So, please don't say that "linear regression doesn't work" . . . that is simply a false statement.
The real issue is what you're trying to do.
If you are trying to develop a methodology that will help you understand the market and be able to make a highly accurate prediction at any point, then you're in for a lot of hard work. There's a great chance that you don't have the know-how, IQ, funding or tools to succeed. You're going to become a statistic.
If, on the other hand, you are TRYING TO MAKE MONEY . . . it doesn't take a lot to do it. Develop a simple method that will earn you $100/day. When you have your hundred bucks, quit for the day. Go do something else. Do that every day. Increase contracts slowly, say one a month over 5-6 months.
By setting yourself up with simple, low expectations, you stay away from all but the highest probability plays. (Of course, this assumes you have the patience and self-discipline to wait.)
Here's an example of a simple way to do this with linear regression:
Wait for the 9 pd ema to be above the 18 pd ema on the 15 min chart. Once it is, look for the same alignment on the 3 min chart. Wait for a pullback. Draw the LR channel. Enter on a breakout of the LR channel. (Don't like the 9-18? Use the 5-35, 5-15, 5-20, or whatever suits you.)
So, please don't say that "linear regression doesn't work" . . . that is simply a false statement.
yes key is confluence. whether that is murrey with fibs. lrc with market profile. but like anything else it comes down to patience.
i like putting all the indexes, up one time frame throw in a setting that matches the range of each individual market. shorter one for er and up to longest for dow and when i get two of the market to show confluence good safe trade. not all major reversals. but a high% of getting small profit and then break even. then the ones that do turn you trade and use the bottome channels as scale out methods.
the best thing i like about them is for me i have found it to be only thing that i can use one time frame and use it from entry to exit. so i am not flipping methods day to day. plus it does not lead to overtrading.
besides trading the open by far the best method i have found especially since linn software auto updates. the ones where you have to constantly redraw ===
i like putting all the indexes, up one time frame throw in a setting that matches the range of each individual market. shorter one for er and up to longest for dow and when i get two of the market to show confluence good safe trade. not all major reversals. but a high% of getting small profit and then break even. then the ones that do turn you trade and use the bottome channels as scale out methods.
the best thing i like about them is for me i have found it to be only thing that i can use one time frame and use it from entry to exit. so i am not flipping methods day to day. plus it does not lead to overtrading.
besides trading the open by far the best method i have found especially since linn software auto updates. the ones where you have to constantly redraw ===