[DcifrThs]

[ QUOTE ]
Agreed. I assume you are developing sample trades for recruiting?

Btw - something I have been wondering about (don't work in trading) - when you think about volatility is there a curve or distribution you have in mind (mathematically)? I get the feeling the normal curve probably isn't used anymore...

[/ QUOTE ]

the sample trades are both for fun and for recruiting. i was asked by a hedge fund owner for trade ideas and gave him a few off the top of my head but would like to get a more systemized view written down to email him.

in terms of the distribution fo price changes, a normal approach is still taken (and i take it) with the proviso that i understand and give a large weight to the possibility of the unexpected and thus err on teh conservative side.

ideally, the distribution would follow a power law of something like mandelbrot's multifractal model of asset returns:

|P(t+dt) - P(t)| ~= Ct(dt)^alpha(t)

where Ct and alpha(t) are the prefactor and Holder exponent respectively. Ct refers to the length of the memory of the process and alpha(t) refers to the degree to which a trend is likely to be continued.

for Ito processes, alpha(t) is 1/2 everywhere. if alpha(t) is > than 1/2, there is less randomness in the sense that a move in one direction is more likely to be followed by a similar directional move. when alpha(t) is < 1/2, a move in one direction is likely to be followed by a move in the opposite direction and thus price changes appear "more random"

the probelm with this methodology is that the estimation of alpha(t) and(to a lesser extent) C(t) is unbelievably crucial. small changes in those variables can lead to huge changes in the results so it is not a very robust method for incorporating into portfolio ocnstruction though there are some improvements that make it better.

it also becomes mathematically difficult.

Barron