Quant applications in finance

Pricing and Risk Management

The price of an asset is supposedly the expected value of its future cashflows minus interest that you will earn for saving your money in Treasury. The challenge lies in figuring out what is 'expected cashflows' in as smart a way as possible. It sounds simple but options and credit derivatives are large pricing challenges. The famed black-scholes model is known for being unable to incorporate fat tails, your large 10% a day movements. It gets more and more complex as financial contracts get customized. For the exact reason, risk managers find it difficult to estimate if their traders are taking on too much risk.

Trading - Arbitrage and Prediction

The verdict has yet to be out if the future can be predicted sufficiently well for economic profits. At least, some people claim to have succeeded. Price prediction models that I have come across include the use of Markov regime switching models in FX prediction, which I implemented and threw away. It is good for monthly, not daily prediction. Neural networks, which is what I am currently working on, is a promising field. Genetic algorithm and swarm intelligence are also some of the newer technologies being touted. Genetic algorithm in another form known as genetic programming can be used for the selection of technical indicators. The game of arbitrage is not easy to enter as it is already a very crowded field. Practitioners currently face the challenge of reducing latency in their applications. The guy with the fastest computer wins.