Moving Average or Linear Regression?

Quote from Albert Cibiades:

I prefer a first order infinite impulse response filter myself (better known as an alpha filter to the unwashed), but then again I am infinitely impulsive. However, I cheerfully defer to the artist formerly known as Abogdan. He is the smartest mathematician here. I had a girlfriend once who had a cubic spline. She fit very nicely.
More...

Mike, err... I mean Albert, pardon me for it has been a long time since I delved into all things DSP, but isn't a first order IIR filter a standard Expo-MA in trader's parlance?
 
Quote from kut2k2:

Sometimes you want lag, sometimes you don't. :cool:
More...
Lag or Leg?

If you can't get Leg, do you use the "Right Hand Rule?" :p

I know I do! :D
 
Quote from programtrader:

Does anybody has experience working with these smoothing techniques, that could point any advantages or disadvantages of using MA's or LR's for:

a) trend direction measurement

b) mean reversion measurement

LR's for the same period seem to be more prone to whipsaws, ence better for the measurement of mean reversion patterns?
More...
First, you have to realize that you're mixing apples and oranges.

LR is a model. You assume the data has a linear form and you "force" the best fit by finding the model parameters that give the least overall estimation error. There is no lag here; once you've found the best model parameters, you can forecast out to eternity if you wish.

MA is a smoother. You assume nothing about the form of the data, you are simply trying to reduce noise without losing too much signal. The best MA will necessarily use a different method of adjustment than the best LR, because there is no a priori form to be fitted.

Good luck with your research.
 
I'm glad you stress the word assume, since any type of statistical modelling has a bunch of rules put in place before going forward. The one assumption I love is the one that "all consumers are rational." NOw we all know that's a farce, but let's make the same assumption about investors "all investors are rational." If you make that assumption, modelling it would be a piece of cake, if not, you get what you have now: stochastics, bollinger bands and the like that use stastical means and parameters that work with the assumption that stock data can be "fitted" well

Quote from kut2k2:

First, you have to realize that you're mixing apples and oranges.

LR is a model. You assume the data has a linear form and you "force" the best fit by finding the model parameters that give the least overall estimation error. There is no lag here; once you've found the best model parameters, you can forecast out to eternity if you wish.

MA is a smoother. You assume nothing about the form of the data, you are simply trying to reduce noise without losing too much signal. The best MA will necessarily use a different method of adjustment than the best LR, because there is no a priori form to be fitted.

Good luck with your research.
More...
 
Quote from Albert Cibiades:

Oh, give it up. ALL estimators lag. The market can change direction faster than a woman can change her mind.
More...

I was about to ask, can anyone give me leading indicator for prices? I'm willing to pay, if it works. :)
 
Quote from Albert Cibiades:

Oh, give it up. ALL estimators lag. The market can change direction faster than a woman can change her mind.
More...
LOL.
Al, did you ever study the effects of a woman's weight/girth with respect to the her "rate of change of mind" :)
or is this pretty much a universal constant (enquiring minds want to know, as mine are - except for the occasional chunkier babe - skinny-as-a-rake-supermodel-looking types with 'interesting' bevioural traits).

PS. Give my regards to your uncle Pericles.
 
Quote from Albert Cibiades:

I prefer a first order infinite impulse response filter myself (better known as an alpha filter to the unwashed), but then again I am infinitely impulsive. However, I cheerfully defer to the artist formerly known as Abogdan. He is the smartest mathematician here. I had a girlfriend once who had a cubic spline. She fit very nicely.
More...
I like your filter.
:cool:
 
I was about to ask, can anyone give me leading indicator for prices? I'm willing to pay, if it works. ------------>
----------------------------------------------------------------------------------

For Day trading, You should have different strategy with different indicators, A ranging day needs its own setup and a trending day requires different system.

Standard deviation systems such as Keltner channel penetration would not work on strong trending days, also requires a very serious back testing and a good money management, but they be very mechanical and profitable.

For heavy trending days you just need to simply go with the crowd, I personally use Jack Hershey's system ( for ES ) IF my entry does not happen to be in RSI and MFI red zone.
 
Quote from madmaxer:

I was about to ask, can anyone give me leading indicator for prices? I'm willing to pay, if it works. ------------>
----------------------------------------------------------------------------------

For Day trading, You should have different strategy with different indicators, A ranging day needs its own setup and a trending day requires different system.

Standard deviation systems such as Keltner channel penetration would not work on strong trending days, also requires a very serious back testing and a good money management, but they be very mechanical and profitable.

For heavy trending days you just need to simply go with the crowd, I personally use Jack Hershey's system ( for ES ) IF my entry does not happen to be in RSI and MFI red zone.
More...
Most helpful!
Please tell us now how to spot:
(1) a ranging day;
(2) a strong trending day;
(3) a heavy trending day.
 
You must log in or register to reply here.
Back
Top