Abstract:

An option is a trading contract in which financial products are exercised. With the rapid growth of options trading volume and trading volume, the calculation of option pricing is getting bigger and bigger, and it is more and more difficult to price options on traditional CPU platforms. The emergence and development of Graphics Processing Unit（GPU） platforms provides a solution to the option pricing calculation. Least Squares Monte Carlo（LSM） algorithm is used to calculate the pricing of one-dimensional and four-dimensional American options. First, the CURAND library is used to generate a large number of random numbers, and then the price change path of the option target is parallelized. Finally, the least squares method and discount pricing are parallelized. In addition, to improve the computational efficiency of the LSM method on the GPU platform, the entire process is optimized. The actual test results show that the acceleration ratios of the one-dimensional and four-dimensional American option pricing on the CPU+GPU are up to 20.275 and 47.538 respectively, and the overall performance of the methods of other literature has been greatly improved.

Key words: GPU, Least Squares Monte Carlo（LSM）, American option pricing, parallel computing

摘要：

期权是以金融产品作为行权品种的交易合约。随着期权交易规模和交易量的迅速增长，期权定价的计算量越来越大，在传统CPU平台上对期权进行定价变得越来越困难。图形处理器（GPU）平台的出现和发展为解决期权定价计算提供了解决方案。在GPU上使用最小二乘蒙特卡罗算法（Least Squares Monte Carlo，LSM）实现了对一维和四维美式期权定价计算：首先利用CURAND库产生大量随机数，然后并行化期权标的价格变化路径，最后对最小二乘法和贴现定价进行并行化。为提高GPU平台上LSM方法的计算效率，对整个过程进行了优化。实际测试结果表明，在CPU+GPU上实现一维和四维美式期权定价相对CPU平台的加速比最高分别达到20.275和47.538，且比其他文献的方法整体性能有较大的提升。

关键词: GPU, 最小二乘蒙特卡罗算法（LSM）, 美式期权定价, 并行计算