Adaptive Moving Averages

ScoobyStoo · Jun 10, 2009

Hi All,

I wonder whether any of you have any interesting opinions on adaptive moving averages.

I need an AMA which will help me accurately identify market turning points. As such I have no interest in the actual value of the AMA (it can overshoot as much as it likes) because I will only be measuring the rate of change of the AMA and using the cross of the zero line as my trigger. Really, I'm only interested in its adaptive noise filtering characteristics and how quickly it reacts.

Obvious candidates for evaluation are the KAMA, MAMA and VIDYA (not bothering with the JMA as I don't need to accurately track price and want access to the algos so I can fiddle around under the hood myself if need be).

Anyone have any pearls of wisdom about these indicators, or indeed any suggestions for others I've missed?

Many thanks in advance.

intradaybill · Jun 10, 2009

Quote from ScoobyStoo:

I need an AMA which will help me accurately identify market turning points.

IMO there is no such thing or possibility, even remotely and when no accuracy is required. All MAs tell you, adaptive or not, and with a time lag is what happened in the past.

AMAs are better in choppy markets but are as good as EMAs in trending markets.

ScoobyStoo · Jun 10, 2009

Quote from intradaybill:

IMO there is no such thing or possibility, even remotely and when no accuracy is required. All MAs tell you, adaptive or not, and with a time lag is what happened in the past.

AMAs are better in choppy markets but are as good as EMAs in trending markets.

Take your point Bill. As I said though...

"I need an AMA which will help me accurately identify market turning points."

I am not relying on the AMA.

I want a time lag as I am evaluating the market both before and after the turn. I am not interested in catching the exact moment of the turn, hence a lagging MA is the correct tool for the job. What I am very interested in is filtering choppy noise to eliminate whipsaw trigger signals.

intradaybill · Jun 10, 2009

Quote from ScoobyStoo:

Take your point Bill. As I said though...

"I need an AMA which will help me accurately identify market turning points."

I am not relying on the AMA.

I want a time lag as I am evaluating the market both before and after the turn. I am not interested in catching the exact moment of the turn, hence a lagging MA is the correct tool for the job. What I am very interested in is filtering choppy noise to eliminate whipsaw trigger signals.

Yep, language can be tricky sometimes. I think an AMA is a good choice for eliminating whipsaw trigger signals depending on your objectives of course. Some people profit from such whipsaws, others, like trend followers, try to avoid them.

rickf · Jun 10, 2009

I had good luck with the MAMA/FAMA combo (the 'MESA' Moving Average) last year --- had to tweak it a bit for various trading styles but it wasn't too bad as far as MAs go.

FWIW saying the only MA on my charts is the 200 sma so rest assured I don't use them regularly.

Still, MESA was pretty handy, check it out.

codetroll · Jun 10, 2009

Not sure if this was the road you were trying to go down, but I'll throw in my thoughts that came into my head randomly.

Maybe you could incorporate/use an LSMA (Least Squares Moving Average) into what your trying to do with perhaps a combination of the MSROC (Moving Slope Rate of Change). Sounds like if you could combine those two you'd have something close to what you want.

Good luck and let us know how your experimenting goes.

-troll

ScoobyStoo · Jun 11, 2009

Quote from intradaybill:

Some people profit from such whipsaws, others, like trend followers, try to avoid them. [/B]

Yep. But the scale I'm looking at makes whipsaws untradeable unless you are a very quick witted local making on the spread. Hence to me it is just noise and I want to filter it.

Thanks for your thoughts.

ScoobyStoo · Jun 11, 2009

Quote from rickf:

I had good luck with the MAMA/FAMA combo (the 'MESA' Moving Average) last year --- had to tweak it a bit for various trading styles but it wasn't too bad as far as MAs go.

FWIW saying the only MA on my charts is the 200 sma so rest assured I don't use them regularly.

Still, MESA was pretty handy, check it out.

Yeah, the MAMA looks interesting. Currently trying to figure out how to tweak the sensitivity of it. Any tips?

intradaybill · Jun 11, 2009

Quote from ScoobyStoo:

Yep. But the scale I'm looking at makes whipsaws untradeable unless you are a very quick witted local making on the spread. Hence to me it is just noise and I want to filter it.

Thanks for your thoughts.

You're probably right. You may want to look into a method of switching from an AMA to a WMA when their difference becomes small. But I have to tell you about my experience, MAs, sooner or later will do the damage.

ScoobyStoo · Jun 11, 2009

Quote from codetroll:

Not sure if this was the road you were trying to go down, but I'll throw in my thoughts that came into my head randomly.

Maybe you could incorporate/use an LSMA (Least Squares Moving Average) into what your trying to do with perhaps a combination of the MSROC (Moving Slope Rate of Change). Sounds like if you could combine those two you'd have something close to what you want.

Good luck and let us know how your experimenting goes.

-troll

Thanks for the help. I've played around with linear regression moving averages before and found them useful. Never come across the MSROC though. Will take a look.