which more advanced methods would you look at ...

M

man

... if you only had the time and skill?

i am currently contemplating on sponsoring some academic work. i would like them to do some initial research on advanced techniques which would take us several weeks each. reason for this is that we are already swamped with development issues and i need to concentrate on topics where i have indication that it could be worth it.

wavelets,
agent based models and
signal processing are now on my list.

i am not so fond of nn, since i am naturally paranoid of overfit and nn still seem to me to be the ultimate fitting candidate. we are already moving towards genetic algorithms, so that is covered. chaos theory is something i love to read about, yet have not indication that it can be applied profitably.

data basis will be index futures, not lower than minute bar level.

anyone having hints for addOns to "the list"?


peace
 
Dynamic time analysis and statistics.
Everything else is just overly complicated for its own sake and will not produce better results.
 
Quote from Walther:

Dynamic time analysis and statistics.
Everything else is just overly complicated for its own sake and will not produce better results.
More...


well, though we do not use too much fancy stuff at the moment i do not see any reason why tough methods should NOT work. and i'd hope to get signals with not too many other people ...
 
Quote from man:

well, though we do not use too much fancy stuff at the moment i do not see any reason why tough methods should NOT work. and i'd hope to get signals with not too many other people ...
More...

Look at it this way, you will hire quant or analyst to work on some weird stuff like wavelets or advanced Fourier. What is his motivation to give you his best results if he gets paid anyway ?
None of those "advanced methods" make 20% per year , everybody knows that . Your best bet is to hire performance based analyst who does some off the wall stuff for your experimental account instead of wasting time on stuff like NN and such.
 
Theoretical physics and models of cognition and perception, this is somewhat related to agent-based modelling which is another pretty exciting field but I believe the wrong approach.
 
Quote from man:

i assume this is supposed to be amusing ...
More...

From coolweb it is amusing....

He is an airhead....he tells us.

For making high velocity money, it is not.

If you do parasitic trading front running smart money, you can count on volume to be a terrific trigger and for optimizing capital movement.

Depends upon how much capital you have to spread around and, thus, how many markets.

Take a look around at your lists and priorities; if you have a subset that involves moving your money velocity to higher ground, then give consideration to front running smart money. There are two keys; volume is one of them for several strategic reasons all unknown to coolweb.

I am assuming you are way beyond the % Walther mentions.
 
Quote from man:

i assume this is supposed to be amusing ...
More...

you asked:
which more advanced methods would you look at ...
What would you expect? You only will get the weirdest crap for answers to your question.

No other way than finding out for yourself by working very hard at it.
 
It's a legitimate question, interesting, and seems well worthy of discussion!

And yet, as nononsense well pointed out, even when one has 'the answer' (and there are many) the road may or may not lead to quintessence or success.

-imho, of course...
 
You must log in or register to reply here.
Back
Top