Monday, August 28, 2006

Enough of Gaussian already

For those who read about the mysterious golden ratio phi, what I am about to write about now would bear some resemblance. While they are completely unrelated, I feel that both are akin in terms of their ability to explain real-world situations. So what exactly am I writing about? Yes, it’s the power-law distribution! The purest form of power-law distribution one can think of is the discrete Pareto distribution. Frequencies of words in longer texts, sizes of human settlements, values of oil reserves in oil fields, sizes of sand particles, sizes of meteorites, numbers of species per genus, areas burnt in forest fires and even sexual contact networks seem to follow this distribution. Amazing! But what interest me the most is that individual stock returns can also somewhat be approximated by this distribution as popularized by Benoît Mandelbrot and Nassim Taleb. I'll write more about this when I have the time.

Friday, August 25, 2006

So you wanna be a financial engineer?








Hilarious comic from Financial Engineering News. Would you have given the same answers?

Models for Malaysia

Its been ages since my last post. Been reading alot about quant models lately and have been thinking to myself; how many Malaysians understand QF?

Models, just how much do you know about them? Seldom do we ever ponder the validity of quantitative models for the fact we all seem to have an "efficient market" chip embedded at the back of our head. Models can't be wrong, if they are, someone would have corrected it instantaneously! I remember reading a joke a couple of months back about an efficient market theorist who refused to pickup a 100 dollar bill on the walkway because he believed that markets were efficient such that opportunities like this would have been short-lived. That being said, outcomes generated from quant models are often treated as magic numbers assets prices should cohere to. The way I see it models will gradually become ineffectual across time and new models will have to be built to accommodate the impacts of new variables. Thankfully, the human mind's ability to capture idiosyncrasies naturally facilitates the creation of complex models.

Building a model from scratch requires identifying concrete homogeneity from a reasonably huge dataset with the level of difficulty very much dependent on the correlation factor. Unfortunately, the place where I'm living the finance community lacks the interest and understanding of quantitative models. Appreciation and the understanding of financial mathematics remains ambiguous as market participants are only focused on the outputs and how monetary gains can be achieved from them. My take on this is that a market will never develop if you do not have the right people participating in it. Every market participant's dream is to have an efficient market but it can only be as good as a dream if we don’t have skilled people constantly building models to price and trade assets. Take for example, how of many of you who are reading this post really understand the role of geometric Brownian motion in developing probabilistic models for the derivatives market? Do leave a comment if you're one of those who do, perhaps you could do something for the Malaysian market.