"Adaptive" EMAs

[newDave]

I have an idea to create a sort of “adaptive” EMA with period being changed depending on some volatility index (e.g. ATR). So this classical formula EMA[k, n] = EMA[k-1, n]+(2/(n+1))•(P-EMA[k-1, n])
gonna be extended with some F(ATR) instead of constant n. As you can guess it could be then used to build more or less standard in rest parts trend following strategies.

What do you think about this? Your experience with that is interesting.
Especially interesting any ideas on how this F(ATR) ( or F(other Volatility index)) should look like.
Maybe some classical books or indicators already known?

 

[newDave]

Just a short update for my investigations...
I tryed the Adaptive MA called VIDYA.
EMA[k, n] = EMA[k-1, n]+(2/(n+1))•(P-EMA[k-1, n])*DX/100

Where DX- is a indacator used as a base for famous ADX.
In original formula Abs(CMO) used instead of DX, but I did not recognise a difference.

Unfortunately I have to say that optimization of my strategy does not show significantly better results when moving from simple EMA to VIDIA.

Any better experience?? Wellcome to share it...