Research On Monte Carlo Option Pricing Algorithm For Integrated Many-core Architecture

With the rapid development of global financial market,option,as a representative financial derivative,is constantly innovating,its pricing problems are increasingly complicated.The traditional Black-Scholes formula has been widely used in the field of option pricing,but it has its own limitations,and does not consider the randomness of volatility in the market.In recent years,with the continuous expansion of the scale of options trading and the strong demand for real-time financial market,the traditional pricing method with complex mathematical calculation model can not meet the demand,the option pricing under the condition of random fluctuation of volatility and how to guarantee real-time pricing are the question that needs further study.In view of the option pricing problem in different scenarios,researchers at home and abroad have proposed various solutions,but there are relatively few researches on option pricing under the condition of random fluctuation of volatility.In addition,many solutions do not take real-time issues into account.The research in this paper is mainly applied to the real-time option pricing scenario under the condition of volatility change.Aiming at the problem that the stochastic volatility model Heston is difficult to solve and high in dimensionality,a Monte Carlo option pricing algorithm is proposed.Firstly,based on the theory of random number generator,various random number algorithms are studied.Then,the price of the underlying asset is discretized by the Euler discretization method,and then calculate the value of each option price path by means of random simulation.For the Monte Carlo algorithm,the stochastic simulation process will produce the variance problem,and the combination of stratified sampling and importance sampling is proposed.The method reduces the variance and improves the calculation accuracy.On the basis of the real-time problem of option pricing,based on the theory of serial random number,a high-performance random number algorithm based on MIC is proposed,and the CPU+MIC collaborative computing is designed in combination with many core architecture equipment.Mode to speed up the simulation process.Improve the efficiency of the algorithm.The simulation experiments show that the Monte Carlo option pricing algorithm proposed in this paper for Heston model converges faster than other methods.The proposed compound variance reduction technique has better variance reduction eff ect than the single variance reduction technique.The proposed high-performance random number algorithm that is based on MIC has an optimal single-thread speedup ratio of 10.682 compared with CPU.Compared with the running time of the Monte Carlo option pricing algorithm on the CPU and MIC platform,the co-calculation model of CPU+MIC proposed in this paper has the best acceleration effect on the algorithm.It simulates 1000,000 times,and the fastest running time is 48.12 s under 128 threads,and the highes t speedup ratio is up to 90.