Re: low-lag smoothing filter fitting problem

[Bob Fulks <bfulks@xxxxxxxxxxxx>]

At 01:29 AM 4/27/2004, DH wrote:

>> I have a problem figuring out exactly how to measure the quality of
>> fit for a low-lag smoothing filter applied to market data.
>
>Join the club. To measure 'quality' you need to know what values a
>'perfect' indicator would output. But, if you know what 'perfect' is,
>you could just write a 'perfect' indicator and trade it. Catch 22.
>Chicken and egg. Good luck, I don't think you're going to find the
>answers you are looking for here.

I often define a lagged moving average as "perfect".

So take some good smooth moving average such as the T3 average, plot it back in time an interval equal to it's lag. This is then a smoothed version of the signal with no lag.

Now compare your filter with that signal. Ideally, you would like to track it exactly, which is, of course, impossible. 

Bob Fulks