Financial compilers

SciPDE™ and SciMC™ are the super-charged financial compiler components of the SciFinance solution. Developed to take the programming out of modeling, SciPDE and SciMC automatically generate C/C++ source code for any partial differential equation (PDE), partial integro-differential equation (PIDE) or stochastic differential equation (SDE). Together, they give you the power, speed and accuracy of PDE methods and the robustness, speed and flexibility of optimized Monte Carlo simulations.

How SciPDE and SciMC work

Starting with ASPEN (Algorithm SPEcification Notation), a very high-level, declarative language that closely matches the way you think about your financial modeling problems, and leveraging an extensive knowledge base about finance, mathematical algorithms, and program optimization, SciPDE and SciMC automatically generate the source code for your derivatives pricing model.

The ASPEN language

ASPEN is a concise, flexible, and extensible language for specifying the derivatives structures that you want to model. Its financial and mathematical constructs allow you to describe concisely the fundamentals of your problem in terms similar to those you would use when describing the problem to a colleague, often with a few simple keywords. From this minimal specification, SciFinance can generate pricing code using its default decisions about numerical methods and code interfaces. Alternatively, you can specify the desired numerical methods and interface choices yourself. For example, you can experiment with different solvers, finite difference schemes, or interpolation methods simply by changing one line in the specification.

SciMC (Monte Carlo) modeling methodology

SciMC translates ASPEN specifications of derivatives structures and stochastic processes into pricing models in C/C++ source code. The methodology employed in the generated option pricing codes is the numerical integration of SDEs for the pertinent processes via a variety of numerical techniques. The simulated paths are used to calculate the discounted expected value of the payoff and intermediate cash flows. The stochastic processes may contain state-dependent parameters and jumps with general probability distributions. Sensitivity functions (i.e., greeks) can be defined for any parameters of the SDEs, the payoff, and path dependency parameters.

SciPDE (Partial Differential Equations) modeling methodology

SciPDE translates ASPEN specifications of derivative structures and partial differential equations into pricing models in C/C++ source code. The methodology employed in SciPDE-generated pricing codes is the numerical solution of PDEs via finite difference techniques. SciPDE can also handle the PIDEs that result from modeling jump diffusion and pure jump processes and many nonlinear PDEs such as those arising from consideration of transaction costs. The user simply specifies any linear PDEs or PIDEs describing the evolution of the option value under the assumed process models; appropriate linear boundary and expiration conditions; contract dates; and the actions (e.g., monitoring of discrete path dependencies or Bermudan exercise) to take place on those dates.