Dilemma in modeling idiosyncratic high-frequency noise

[bluelou]

Why does idiosyncratic noise exist in the first place? What causes the idiosyncracies? Is there any academic research on idiosyncratic noise signatures in high frequency data (of any type) that you would recommend? I'm open to your suggestions.

Background:
I use a model based on ideas derived from statistical physics to trade FX and other futures on an intraday basis. The only inputs to the model are price and bar count.

I'm refering to idiosyncratic noise, not signal. That is, I've detrended the data series so that all that is remaining is a noise residual. I believe that I'm finding that FX pairs (and other futures instruments) can have unique and persistent high-frequency noise signatures.

 

[IanMacQuaide]

Truth is ALWAYS stranger than Fiction.

 

[psytrade]

It's the sound of the Market talking to itself

 

[MGJ]

A dilemma is a forced choice between two unpalatable alternatives. What you describe, is not a dilemma. Try other words. Express yourself.

 

[bluelou]

MGJ,
The dilemma is that either I'm wrong about the embedded structure in the noise or there is an embedded structure and I'm the only one who can see it. Neither choice is acceptable to me. Does that help you?

FWIW, since I couldn't find the answer to my question on my own I thought I'd give a retail chat site like ET a try. But, then I knew I ran the chance that people like you would waste my time. Congrats, you are part of a dilemma, too.

-bluelou

P.S. I saw from your post a few days ago that your favorite Fibonacci numbers are 1 and 2. That's nice.

 

[kinggyppo]

http://www.mesasoftware.com/

 

[Tums]

Quote from bluelou:

...I'm refering to idiosyncratic noise, not signal. That is, I've detrended the data series so that all that is remaining is a noise residual...

What does it mean by "Detrend"?

 

[Hook N. Sinker]

Quote from bluelou:

I've detrended the data series so that all that is remaining is a noise residual.

How do you know that all that remains is noise?

 

[trader3cnd]

Quote from bluelou:

I'm refering to idiosyncratic noise, not signal. That is, I've detrended the data series so that all that is remaining is a noise residual.

When you detrend a data series what remains is NOT necessarily noise. It can be a regular sine wave or any other deterministic path.

On the other hand, what makes you think that the trend is not also part of the noise?

I think you are looking at the market from a signal processing perspective and that is not useful at all.

 

[good one]

To me the problem with DSP and the markets is that any one trade is part of the signal, the only real noise is slippage and transaction costs. While maybe this type of model can line up occasionally with the markets, It strikes me that it can not do that better than chance because you know from the get go your treating alot of the signal as noise when it is not.
Really though you would be better off asking this on nuclear phynance or wilmott.