DSP Acceleration for Dynamic Financial Models

Abstract

We present an extensive dynamic financial model that encompasses most models used today in finance and economics. We show that this model is a good match to the capabilities of DSP chips. Particularly, DSP is able to perform the high-speed Monte Carlo simulations that are required to solve many large-scale, intractable financial problems. By simulating a sufficiently large number of future scenarios, DSP chips can rapidly achieve a good approximation of the probable future joint probability distribution function of modeled variables. This probability distribution can be used for the valuation of financial derivatives, computing value at risk, studying macroeconomic policy decisions, and many other purposes. DSP enables such simulations to be faster, cooler, greener, and cheaper than ever before.