Branches of Optimization

trippel · August 21st, 2009, 6:47 pm

I wonder what branches of optimization are the most useful for QF. Of course there are the classical branches LP, IP/MIP, NLP, where the latter can be subdivided into many useful branches like QP, Mixed-Integer-NLP, SDP, ...Do you have any opinions on these? What about hybrids such as Stochastic Optimization? I have the "opportunity" to do my diploma thesis (german degree which is comparable to Msc) in the field of Stochastic Optimization, i.e. Portfolio Optimization in Levy Markets. Is it worth it?

rusti999 · August 22nd, 2009, 2:34 am

You might want to check this book if you haven't already:Optimization Methods in Finance

trippel · August 22nd, 2009, 7:09 am

Thanks. In fact I wasn't aware of this book. But also your personal experience is welcome. I checked the index of the book: From this I can take that the methods are explained in its own right, i.e. IP and NLP are discussed apart from each other. But quite often, one finds Mixed-Integer Non-Linear Programs. However, there are still standard methods to solve, like Early Branching.

Anthis · August 22nd, 2009, 7:33 am

Ziemba

trippel · August 22nd, 2009, 8:41 am

Nice intro, thanks. One of my colleges was able to do a little better optimizing at a particular problem, yielding an increase in profit of about 50 M!!! I am stunned about the impact of successful optimization in finance. Usually somebody working on optimization does not get much attention until it is about form optimization and his sailing yacht wins a competition .

KackToodles · August 22nd, 2009, 8:06 pm

these methods sound impressive and fancy to the novice. but none of them work well because financial markets obey their own "behavior" rules -- not the rules of optimization math. that's why insurance companies like AIG go bankrupt.