Kalman Filter

[377OHMS]

Quote from Arthur Deco:

Jackoff, what a complete crock of shit. I don't know if YOU ever designed Kalman's, but I have. If you understood them like I do, you'd know that there is NO possible physical model for markets. The markets are like women. They can displace instantly. So you need an infinite number of derivatives to model it. Also, there is NO noise in the markets. It's all signal. Dipshit.

Excellent.

The state vectors would have to contain high order derivatives which are noisy. That is exactly correct. Kalman filters are good for physical dynamics because those can generally be covered by position and its first and sometimes second derivatives velocity and acceleration. Market data contains instantaneous displacements that would require third and fourth derivatives jerk and yank. Those are incredibly noisy and basically unusable.

 

[panzerman]

DECO & 377OHMS,

Do you guys have an opinion on using the MESA (All Poles) method to find and trade the dominate cycle? Thanks.

 

[377OHMS]

Quote from panzerman:

DECO & 377OHMS,

Do you guys have an opinion on using the MESA (All Poles) method to find and trade the dominate cycle? Thanks.

I'm not familiar with that.

 

[Equalizer]

Quote from Arthur Deco:

......The only thing that works is weird hypothesis testing by statistical approaches.

Now look what you've done!

 

[Arthur Deco]

Quote from 377OHMS:

Excellent.

The state vectors would have to contain high order derivatives which are noisy. That is exactly correct. Kalman filters are good for physical dynamics because those can generally be covered by position and its first and sometimes second derivatives velocity and acceleration. Market data contains instantaneous displacements that would require third and fourth derivatives jerk and yank. Those are incredibly noisy and basically unusable.

See Jackoff? Jerk and yank. 377, I would address you by your resistor color code, but I have forgotten how. Let's see... "Black boys rape our young girls, but Violet gives willingly, get some now!" It's been so long since I designed anything physical that I couldn't design a voltage divider. Algorithms and software are so much better behaved. Your explanation was very eloquent, and what I meant to say. I would only add that there is no system noise and no measurement noise, so a linear Kalman would devolve to however many derivatives you needed to approximate price action.

 

[Arthur Deco]

Quote from panzerman:

DECO & 377OHMS,

Do you guys have an opinion on using the MESA (All Poles) method to find and trade the dominate cycle? Thanks.

All price cycle theory, of whatever stripe, is bullshit. Price action is all about generating volume for accumulation or distribution by the big dogs, and support and resistance along the way.

 

[Arthur Deco]

Quote from Equalizer:

Now look what you've done!

Oooops! Fortunately the tools to do it are painful to even think about, much less implement.

 

[pspr]

This is the metastock code for an indicator named the Kalman filter. I'm sure Art will rip the code as not even close. Anyway, the resulting indicator does not appear to have any usefulness.

{Kalman Filter}
{Advantage Systems: www.advantage.online.pl}
PRICEE:=C;
Smooth:=.13785*(2*PriceE - Ref(PriceE,-1)) +
.0007*(2*Ref(PriceE,-1) - Ref(PriceE,-2)) +
.13785*(2*Ref(PriceE,-2) -Ref(PriceE,-3)) +
1.2103*PREV - .4867*Ref(PREV,-1);
SMOOTH;

 

[Arthur Deco]

Quote from pspr:

This is the metastock code for an indicator named the Kalman filter. I'm sure Art will rip the code as not even close. Anyway, the resulting indicator does not appear to have any usefulness.

{Kalman Filter}
{Advantage Systems: www.advantage.online.pl}
PRICEE:=C;
Smooth:=.13785*(2*PriceE - Ref(PriceE,-1)) +
.0007*(2*Ref(PriceE,-1) - Ref(PriceE,-2)) +
.13785*(2*Ref(PriceE,-2) -Ref(PriceE,-3)) +
1.2103*PREV - .4867*Ref(PREV,-1);
SMOOTH;

An odd looking beast. All odball filters are a solution to some problem, you just don't know what that problem is.

 

[panzerman]

Here is a good reference article. As has been mentioned, there is nothing particularly magical about Kalman filters or any other filter in and of themselves.