PricingCatalyst

Short reaction times in agile markets, large volumes and low latency applications such as algorithmic trading systems require fast derivative pricing algorithms. PricingCatalyst is a GPU accelerated financial pricing library, which is designed highest performance and large throughput.

For large projects we also offer source code and a complete cross platform development and build system to accelerate in-house GPU software engineering and prevent vendor lock-in.

Features

  • Ultra fast high quality PDE solvers
  • Supporting all kind of option types, such as
  • European options, barrier, multi-barrier, time –window barrier
  • American, and Bermudan exercise
  • Discrete dividends
  • Build new structures by combining payoff, exercise and barrier features
  • Sophisticated grid building infrastructure to avoid oscillatory effects
  • Multiple levels of parallelism, fully exploiting modern GPU hardware
  • Parallel execution on multiple GPUs
  • 2000 and more option price calculations per second on a single GPU
  • High performance Monte Carlo simulation framework
  • Fast high quality pseudo-random and Sobol quasi random number generators
  • GPU path generators for local volatility and stochastic volatility models exploiting specific GPU memory features such as texture memory
  • GPU payoff templates for various single underlying and basket options
  • Run 100’000 path to price an equity basket products with 4 underlyings in less than 30 ms.
  • Generic Greek engine for sensitivity calculation of multi-underlying products
  • GPU accelerated calibration such as fast parallel local volatility calibration
  • GPU accelerated curve and surface library for general purpose low level data processing
  • Scripting support through Python bindings