This paper proposes a hardware implementation for pricing Collateralized Debt Obligations using a recursive analytical method. A novel approach using FIFOs for storage is implemented for the recursive convolution, addressing the main drawback of the analytical approach. The FIFO-based convolution approach is compared against two different other convolution approaches based on the Fast Fourier Transform and the Output Side Algorithm. The FIFO-based convolution outperforms the other two approaches and also results in significant reduction in memory usage.
The CDO core designed with the FIFO-based convolution method is implemented and tested on a Virtex-5 FPGA. Compared against a C implementation running on a 2.8GHz Intel Processor, the CDO core shows a 41-fold speed up. A comparison against a Monte Carlo based hardware implementation for structured instruments yields mixed results. The CDO core performs well against high accuracy Monte Carlo models but it is outperformed when fewer scenarios are considered for the Monte Carlo model.